SG24-4892-00

DFSMS/MVS V1R4 Technical Guide

June 1997
IBML
SG24-4892-00
International Technical Support Organization

DFSMS/MVS V1R4 Technical Guide

June 1997
 Take Note!

Before using this information and the product it supports, be sure to read the general information in
Appendix B, “Special Notices” on page 143.

First Edition (June 1997)

This edition applies to Version 1, Release 4 of DFSMS/MVS, Program Number 5695-DF1 for use with the MVS/ESA
platform and OS/390 operating system.

Warning

This book is based on a pre-GA version of a product and may not apply when the product becomes generally
available. It is recommended that, when the product becomes generally available, you destroy all copies of
this version of the book that you have in your possession.

Comments may be addressed to:

IBM Corporation, International Technical Support Organization
Dept. QXXE Building 80-E2
650 Harry Road
San Jose, California 95120-6099

When you send information to IBM, you grant IBM a non-exclusive right to use or distribute the information in any
way it believes appropriate without incurring any obligation to you.

 Copyright International Business Machines Corporation 1997. All rights reserved.

Note to U.S. Government Users — Documentation related to restricted rights — Use, duplication or disclosure is
subject to restrictions set forth in GSA ADP Schedule Contract with IBM Corp.
Contents

Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
The Team That Wrote This Redbook . . . . . . . . . . . . . . . . . . . . . . . . . xi
Comments Welcome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

Chapter 1. Introduction to DFSMS/MVS Version 1 Release 4 . . . . . . . . . . 1

1.1 DFSMS/MVS Release Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Functional Enhancements . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Usability Enhancements in DFSMS/MVS V1R4 . . . . . . . . . . . . . . 4

Chapter 2. DFSMSdfp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Space Allocation Failure Reduction . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Allocations Affected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Implementation Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.3 Volume Selection Failure Recovery . . . . . . . . . . . . . . . . . . . . 8
2.1.4 New Allocations (VSAM and non-VSAM) . . . . . . . . . . . . . . . . . 9
2.1.5 Extents on New Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.6 Non-VSAM Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.7 VSAM Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.8 Exceptions to Using Volume Selection Failure Recovery . . . . . . . . 10
2.1.9 ISMF Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.10 ISMF Message Changes . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.11 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 SAM Tailored Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.1 Tailored Compression and Generic Compression . . . . . . . . . . . . 13
2.2.2 Sampling the Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.3 Enablement of Tailored Compression . . . . . . . . . . . . . . . . . . . 15
2.2.4 Wellness Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.5 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.6 Functional Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.7 Migration and Coexistence Considerations . . . . . . . . . . . . . . . . 18
2.3 O/C/EOV Serviceability Enhancements . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 New Trace Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.2 Enabling IFGOCETR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.3 IFGOCETR Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.4 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Enhanced Protection of Checkpointed Sequential DASD Data Sets . . . . 22
2.5 Optical Access Method SMF Recording Enhancement . . . . . . . . . . . . 29
2.5.1 Start Time and End Time Accuracy . . . . . . . . . . . . . . . . . . . . 30
2.6 Program Management 3 (PM3) . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.6.1 Program Management - Background . . . . . . . . . . . . . . . . . . . 32
2.6.2 PM3 Enhancements for DFSMS/MVS V1R4 . . . . . . . . . . . . . . . . 33

Chapter 3. DFSMSdfp VSAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.1 VSAM System-Managed Buffering . . . . . . . . . . . . . . . . . . . . . . . . 35
3.1.1 Record Access Bias Specification . . . . . . . . . . . . . . . . . . . . . 36
3.1.2 ISMF Data Class Panels . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.1.3 Buffers Used for Direct Optimization . . . . . . . . . . . . . . . . . . . . 37
3.1.4 Location of Buffers and Control Blocks . . . . . . . . . . . . . . . . . . 38
3.1.5 LSR Resource Pools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

 Copyright IBM Corp. 1997 iii

 3.2 VSAM Load Enhancements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.1 Invocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.2 Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3 Update VSAM Last Reference Date at CLOSE . . . . . . . . . . . . . . . . . 41
3.4 VSAM RLS KSDS Extended Addressability . . . . . . . . . . . . . . . . . . . 41
3.4.1 Objectives and User Requirements . . . . . . . . . . . . . . . . . . . . 42
3.4.2 Functional Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.4.3 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.4 Software Prerequisites . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.5 Configuration Specifications . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5 Data Class Supports More VSAM Attributes . . . . . . . . . . . . . . . . . . 44
3.5.1 User Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5.2 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.6 DCOLLECT Enhancements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.6.1 DCOLLECT OUTPUT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.6.2 DCOLLECT Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Chapter 4. DFSMShsm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1 Duplex Tape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1.1 Automatic TAPECOPY Scheduling . . . . . . . . . . . . . . . . . . . . . 52
4.1.2 Supported DFSMShsm Functions . . . . . . . . . . . . . . . . . . . . . . 52
4.1.3 Invocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1.4 Initial Tape Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1.5 TAPECOPY Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1.6 Tape Exception Processing . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.1.7 AUDIT Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1.8 LIST Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1.9 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1.10 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Alter without Recall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 New IDCAMS Interface with Catalog . . . . . . . . . . . . . . . . . . . . 56
4.2.2 New Catalog Notification of DFSMShsm . . . . . . . . . . . . . . . . . . 56
4.2.3 DFSMShsm Processing Considerations . . . . . . . . . . . . . . . . . . 56
4.2.4 DFSMShsm Error Processing . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3 ABARS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.1 Output File Stacking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.2 Up to 64 Concurrent Requests . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.3 Invocation of ARCBEEXT Extended to DFSMSdss Processing . . . . . 58
4.3.4 GDG Base Name in ALLOCATE Statement . . . . . . . . . . . . . . . . 59
4.3.5 Automatic Delete of ABARS Activity Log during Roll-Off Processing 59
4.3.6 CPU Time for Aggregate Processing in WWFSR . . . . . . . . . . . . . 59
4.3.7 TGTGDS and OPTIMIZE Keyword Externalized . . . . . . . . . . . . . 60
4.3.8 ISMF Changes for ABARS Accounting Information . . . . . . . . . . . 61
4.3.9 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4 CDS Record Level Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.4.1 Invocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.4.2 CDS Access in Record Level Sharing Mode . . . . . . . . . . . . . . . 63
4.4.3 QUERY CONTROLDATASETS Command . . . . . . . . . . . . . . . . . 63
4.4.4 Multicluster CDSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.4.5 CDS Creation or Redefinition . . . . . . . . . . . . . . . . . . . . . . . . 63
4.4.6 DCOLLECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.4.7 ARCIMPRT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.4.8 CDS Backup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.4.9 Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.4.10 Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

iv DFSMS/MVS V1R4 Technical Guide

4.5 DFSMShsm and DFSMSrmm Interface . . . . . . . . . . . . . . . . . . . . . 67
4.5.1 Current Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.5.2 New Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.5.3 Running DFSMShsm with DFSMSrmm . . . . . . . . . . . . . . . . . . . 67
4.5.4 EDGDFHSM Program Interface . . . . . . . . . . . . . . . . . . . . . . . 67
4.5.5 EDGTVEXT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.6 Dump Analysis Elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.6.1 Setup Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.6.2 Programming Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.6.3 Dumps Not Eligible . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.6.4 DFSMShsm Processing in the Primary Space . . . . . . . . . . . . . . 70
4.6.5 DFSMShsm Processing in the ABARS Secondary Address Space . . 70
4.7 Serviceability Support Items . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.7.1 Four New SMF Functional Statistical Record Types . . . . . . . . . . . 70
4.7.2 Aliases for DFSMShsm Keywords . . . . . . . . . . . . . . . . . . . . . 71
4.7.3 AUDIT DATASETCONTROLS . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.7.4 DFSMS/MVS Optimizer HSM Monitor/Tuner . . . . . . . . . . . . . . . 72

Chapter 5. DFSMSdss Multivolume Selection Enhancements . . . . . . . . . . 73

5.1 ALLMULTI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.2 SELECTMULTI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.2.1 SELECTMULTI for DUMP and COPY . . . . . . . . . . . . . . . . . . . . 75
5.2.2 SELECTMULTI for CONVERTV . . . . . . . . . . . . . . . . . . . . . . . . 75
5.2.3 Using the ADRUIXIT Exit . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2.4 Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2.5 Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2.6 New Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2.7 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.2.8 Support Use Information . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.2.9 Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.2.10 Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Chapter 6. DFSMSrmm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.1 Journal Usage Threshold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.1.1 Update PARMLIB Member EDGRMMxx . . . . . . . . . . . . . . . . . . 79
6.1.2 Automating Control Data Set Backup and Journal Clearing . . . . . . 80
6.1.3 Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.1.4 TSO Subcommand Variables by Name . . . . . . . . . . . . . . . . . . 82
6.1.5 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2 Nonintrusive Backup of the CDS . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2.1 Functional Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.2.2 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.2.3 Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3 Problem Determination Aid Trace . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3.1 Functional Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3.2 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.3.3 Support Use Information . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.3.4 Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.4 Support for DFSMShsm Alternate Tape Processing . . . . . . . . . . . . . 88
6.5 Recognition of External Data Managers . . . . . . . . . . . . . . . . . . . . 89
6.6 DFSMSrmm Inventory Management Trial Runs . . . . . . . . . . . . . . . . 90
6.6.1 Inventory Management Processing . . . . . . . . . . . . . . . . . . . . 91
6.6.2 Inventory Management VRSEL Processing . . . . . . . . . . . . . . . . 92
6.7 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.7.1 New EXEC Parameters for EDGHSKP parameters . . . . . . . . . . . . 93

Contents v
 6.7.2 New PARMLIB Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.7.3 New and Changed Messages . . . . . . . . . . . . . . . . . . . . . . . . 94

Chapter 7. Distributed Data Access . . . . . . . . . . . . . . . . . . . . . . . . . 95

7.1 Network File System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.2 DFSMS/MVS Network File System . . . . . . . . . . . . . . . . . . . . . . . . 95
7.2.1 DFSMS/MVS NFS Client Function . . . . . . . . . . . . . . . . . . . . . 96
7.2.2 OpenEdition Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.2.3 DFSMS/MVS NFS Client Commands . . . . . . . . . . . . . . . . . . . . 99
7.2.4 DFSMS/MVS NFS Server Enhancements . . . . . . . . . . . . . . . . 100
7.3 DFSMS/MVS Distributed FileManager . . . . . . . . . . . . . . . . . . . . 101
7.3.1 Distributed FileManager DataAgent . . . . . . . . . . . . . . . . . . . 102
7.3.2 DFM DataAgent Invocation . . . . . . . . . . . . . . . . . . . . . . . . 105
7.3.3 DataAgent Processing (DFM Target) . . . . . . . . . . . . . . . . . . . 108

Chapter 8. DFSMS/MVS Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . 115

8.1.1 Input Data for the Optimizer . . . . . . . . . . . . . . . . . . . . . . . . 115
8.1.2 Performance Analyzer . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.1.3 Management Class Analyzer . . . . . . . . . . . . . . . . . . . . . . . 117
8.1.4 Charting Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.1.5 HSM Monitor/Tuner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

Chapter 9. NaviQuest and Catalog Search Interface . . . . . . . . . . . . . . 119

9.1 NaviQuest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
9.1.1 Enchanced ACS Management Option Panels . . . . . . . . . . . . . 120
9.1.2 NaviQuest Help Panels . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.1.3 Migration and Coexistence . . . . . . . . . . . . . . . . . . . . . . . . 131
9.2 Catalog Search Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.2.2 Invocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
9.2.3 The Parameter List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
9.2.4 Work Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
9.2.5 CSI Sample Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

Appendix A. Parallel Sysplex and VSAM Record Level Sharing . . . . . . . 139

A.1 Overview of Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . 139
A.2 Overview of VSAM Record Level Sharing . . . . . . . . . . . . . . . . . . 139
A.3 VSAM RLS Locking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

Appendix B. Special Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

Appendix C. Related Publications . . . . . . . . . . . . . . . . . . . . . . . . . 145

C.1 International Technical Support Organization Publications . . . . . . . . 145
C.2 Redbooks on CD-ROMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
C.3 Other Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

How to Get ITSO Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

How IBM Employees Can Get ITSO Redbooks . . . . . . . . . . . . . . . . . . 147
How Customers Can Get ITSO Redbooks . . . . . . . . . . . . . . . . . . . . . 148
IBM Redbook Order Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

vi DFSMS/MVS V1R4 Technical Guide

ITSO Redbook Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

Contents vii
viii DFSMS/MVS V1R4 Technical Guide
Figures

1. Tailored Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2. Wellness Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3. Generic and Tailored Compression Measurements . . . . . . . . . . . . 16
4. Dictionary Token . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5. VSAM Loading Enhancements . . . . . . . . . . . . . . . . . . . . . . . . . 40
6. VSAM Last Reference Date at CLOSE . . . . . . . . . . . . . . . . . . . . . 42
7. VSAM Data Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8. DCOLLECT Command Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . 49
9. Problems with ALLMULTI . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
10. Journal Usage Threshold . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
11. CDS and Journal Backup and Clearing . . . . . . . . . . . . . . . . . . . . 81
12. DFSMSrmm REXX Variables for the Journal . . . . . . . . . . . . . . . . . 82
13. DSSOPT DD Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
14. DFSMSrmm Problem Determination Aid . . . . . . . . . . . . . . . . . . . 86
15. EDGUX100 Installation Exit . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
16. Inventory Management Trial Runs . . . . . . . . . . . . . . . . . . . . . . . 92
17. DFSMS/MVS NFS client feature . . . . . . . . . . . . . . . . . . . . . . . . 96
18. The DFSMS/MVS DFM DataAgent. . . . . . . . . . . . . . . . . . . . . . . 102
19. Elements of the DFSMS/MVS Optimizer . . . . . . . . . . . . . . . . . . 116
20. ISMF Primary Option Menu . . . . . . . . . . . . . . . . . . . . . . . . . . 121
21. DFSMSdfp NaviQuest Component Primary Option Menu . . . . . . . . 121
22. Test Case Generation Selection Menu . . . . . . . . . . . . . . . . . . . 122
23. Test Case Generator from Saved ISMF List Entry Panel . . . . . . . . . 123
24. ACS Test Listings Comparison Panel . . . . . . . . . . . . . . . . . . . . 124
25. Enhanced ACS Test Listing Entry Panel . . . . . . . . . . . . . . . . . . . 125
26. Test Case Update with Test Results Entry Panel . . . . . . . . . . . . . 126
27. Batch Testing/Configuration Management Selection Menu . . . . . . . 127
28. Saved ISMF List Operations Batch Samples Selection Menu . . . . . . 127
29. Configuration Changes Batch Samples Selection Menu . . . . . . . . . 128
30. First Edit Screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
31. Second Edit Screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
32. Catalog Search Interface Invocation Structure (Parameter List) . . . . 133
33. The Catalog Search Interface Work Area . . . . . . . . . . . . . . . . . . 137

 Copyright IBM Corp. 1997 ix

x DFSMS/MVS V1R4 Technical Guide
Preface

This redbook introduces the latest version of DFSMS/MVS. It provides a

technical overview of the functions and enhancements of DFSMS/MVS V1R4.

DFSMS/MVS V1R4 improves system and data availability, increases productivity,

offers performance enhancements, and fulfills several important customer
requirements.

The first chapter provides an overview, by functional component, of the new

functions and enhancements in DFSMS/MVS V1R4. Each subsequent chapter is
dedicated to a component of DFSMS/MVS. Installation, migration, and
coexistence considerations with releases of DFSMS/MVS previous to V1R4 are
provided.

Some of the new features of DFSMS/MVS V1R4 covered in this redbook include:
• Reducing space-related outages
• DFSMShsm Duplex Tape, ABARS, and VSAM RLS CDS processing
enhancements
• DFSMSrmm trial-run capability
• System-managed buffering and load enhancements of VSAM Extended
Format KSDSs
• SAM compression improvements
• New OAM SMF records
• Improved access to enterprise data through the DFM/MVS data agent
• Catalog Search Interface
• Batch ACS testing improvements

This redbook is written for experienced storage administrators, systems

programmers, and other technical professionals who require a technical update
of the new features and enhancements in this release of DFSMS/MVS.

The Team That Wrote This Redbook

This redbook was produced by a team of specialists from around the world
working at the International Technical Support Organization San Jose Center.

Mary Lovelace is a Software Engineer at the International Technical Support

Organization, San Jose Center. Before joining the ITSO, Mary worked in the
Education Support Center in Poughkeepsie, New York. She has more than 18
years of experience with IBM in large systems and storage product education,
system engineering and consultancy, marketing support, and system
programming.

Peter Zerbini is an Adviser in Germany. He has worked at IBM for 24 years and
has 12 years of experience in storage software and storage management. His
areas of expertise include DFSMS/MVS, DFSMS Optimizer, DASD, tape, and
optical. Peter is currently the Marketing Support and Last Level specialist for
DFSMS/MVS, DFSMS Optimizer, and optical in Germany.

 Copyright IBM Corp. 1997 xi

 Geoff Littlewood is a Senior Systems Specialist in the United Kingdom. He has
18 years of experience in storage and storage management. Geoff holds an
MSC in Mathematics and Computing Science from Sheffield University in the
United Kingdom. His areas of expertise include DASD, tape, and SMS. Geoff is
currently the Marketing and Technical Support specialist for DFSMS/MVS and
storage management in the United Kingdom.

Thanks to the following people for their invaluable contributions to this project:

Ed Baker
Storage Systems Division - Tucson

Charlie Burger
Advanced Technical Support Organization - San Jose

Cecilia Carranza Lewis

Storage Systems Division - San Jose

Jerry Codde
Storage Systems Division - San Jose

Ed Daray
Storage Systems Division - San Jose

Scott Drummond
DFSMS Brand Manager - San Jose

Tina Dunton
Storage Systems Division - San Jose

Nadine Hart
Storage Systems Division - San Jose

Bob Kern
Storage Systems Division - Tucson

Ron Kern
Storage Systems Division - Tucson

Larry Law
Storage Systems Division - San Jose

William Nettles
Storage Systems Division - San Jose

Deborah Norberg
Storage Systems Division - San Jose

Tony Pearson
Storage Systems Division - Tucson

Jerry Pence
Storage Systems Division - Tucson

Savur Rao
Storage Systems Division - San Jose

xii DFSMS/MVS V1R4 Technical Guide

 Dave Thompson
Storage Systems Division - San Jose

Carmen Yep
Storage Systems Division - San Jose

Joy Nakamura
Storage Systems Division - San Jose

Ron Ward
Storage Systems Division - Tucson

Taylor Winship
Storage Systems Division - San Jose

Mike Wood
Sortware Services - IBM UK

Thanks are also due to the many other collaborators and reviewers who
contributed to this book. I especially want to acknowledge our technical editor,
Maggie Cutler, whose work substantially improved the quality of this book.

Comments Welcome
Your comments are important to us!

We want our redbooks to be as helpful as possible. Please send us your

comments about this or other redbooks in one of the following ways:
• Fax the evaluation form found in ″ITSO Redbook Evaluation″ on page 159 to
the fax number shown on the form.
• Use the electronic evaluation form found on the Redbooks Home Pages at
the following URLs:
For Internet users http://www.redbooks.ibm.com
For IBM Intranet users http://w3.itso.ibm.com/redbooks
• Send us a note at the following address:
[email protected]

Preface xiii
xiv DFSMS/MVS V1R4 Technical Guide
Chapter 1. Introduction to DFSMS/MVS Version 1 Release 4

DFSMS/MVS provides and enhances functions formerly provided by MVS/DFP,

Data Facility Data Set Services (DFDSS), and the Data Facility Hierarchical
Storage Manager (DFHSM). DFSMS/MVS is easier to install and order than
those earlier offerings, as it eliminates the need to install multiple products.
DFSMS/MVS improves on earlier offerings by adding support for new functions.

DFSMS/MVS is a licensed program that delivers four MVS storage management

functional components as a single, integrated software package:
• DFSMSdfp - Provides storage, data, program, and device management
functions
• DFSMSdss - Provides data movement, copy, backup, and space management
functions
• DFSMShsm - Provides backup, recovery, migration, and space management
functions
• DFSMSrmm - Provides management functions for removable media such as
tape cartridges and 3420 reels

DFSMS/MVS V1R4 expands on the DFSMS/MVS product by providing new

enhancements and function. The subsequent chapters in this book describe the
functions of and enhancements to DFSMS/MVS provided in this release, with
each chapter being devoted to a functional area. Thus those readers with an
interest in DFSMShsm or in VSAM can go directly to a particular chapter to see
what has been included or enhanced for that specific functional area. This book
is intended for those who are familiar with DFSMS/MVS and its components.

1.1 DFSMS/MVS Release Focus

DFSMS/MVS provides storage management, data access, device support,
program management, and distributed data access for the MVS/ESA platform.
DFSMS/MVS V1R4 improves system and data availability, performance, and
productivity and fulfills important customer requirements.

What has become apparent since DFSMS/MVS was first delivered is that many of
the most significant functional enhancements can be realized only if the data is
system-managed storage (SMS)-managed. That is true of this release of
DFSMS/MVS as well, as you will discover as you read this book.

Follow-on releases of DFSMS/MVS are likely to continue this trend of delivering

new function and enhancements for SMS-managed data.

Most of the significant benefits of this release of DFSMS/MVS will be available

only to customers who have invested some time and effort in implementing SMS.
For those customers who have not yet made this step, the DFSMSdfp NaviQuest
component functionality should make it easier than ever before to migrate to
SMS, particularly if the DFSMS Fast Implementation Techniques methodology is
used.

 Copyright IBM Corp. 1997 1

1.1.1 Functional Enhancements
DFSMS/MVS V1R4 includes the following new functions and enhancements:
• DFSMSdfp
− Space allocation failure reduction
This enhancement reduces out-of-space conditions for SMS-managed
data sets for new volume processing only. DADSM may use more than
five extents to allocate the requested space quantity. Two new DATA
CLASS definition parameters can be used to influence the allocation and
extension of data sets (to new volumes) in such a manner that
allocations that might have failed for lack of space may succeed.
− Sequential Access Method tailored compression
SAM tailored compression is an enhancement to the DFSMS
compression feature that was introduced in DFSMS/MVS 1.2.0. It creates
dictionaries tailored for the data set that are embedded in the data set.
The primary purpose of tailored compression is to significantly improve
the compression ratio for SAM (physical sequential) data set
organizations. This technique improves the compression ratios over the
generic dictionary building block (DBB) compression techniques.
− Open/close/end of volume (O/C/EOV) serviceability improvements
Many serviceability improvements have been added in this release of
DFSMS/MVS.
− Enhanced protection of checkpointed data sets
Changes have been made to DFSMSdfp so that an MVS physical
sequential (PS) checkpointed data set can be distinguished from an IMS
GSAM checkpointed data set. DFSMSdss operations that can make PS
checkpointed data sets unusable for restart are performed only when
specifically requested.
− Object Access Method (OAM) SMF recording
A new SMF record type 85 is provided for OAM object access. The SMF
record type 85 supports subtypes, and there is a subtype for almost
every activity performed by OAM.
− Program Management 3 (PM3)
This release of PM will include the following enhancements:
- A conversion function added to the Binder to process extended
object (XOBJ) structures built by the C/C++ compilers
- The Binder and Loader will be enhanced to support OS/390 dynamic
linking and dynamic load libraries (DLLs)
- Support for the two above enhancements in OpenEdition and the C89
shell command
- Loader enhancements to support DLLs and deferred loading, as well
as the OS/390 R4 Dynamic LPA functions.
• Virtual Storage Access Method (VSAM)
− System-managed buffering
VSAM system-managed buffering allows VSAM to optimize batch
programs by determining the number of data and index buffers as well

2 DFSMS/MVS V1R4 Technical Guide

 as the type of buffer management—sequential or direct optimized, or
sequential or direct weighted.
− VSAM load enhancements
When system-managed buffering is allowed, the performance of loading
an extended format key-sequenced data set (KSDS) can be improved by
reducing the number of I/Os required to write the data.
− Update last reference date at close
The last reference date will now be updated in the VTOC on the first
volume for the base data component when the data set is closed if the
date at close is different from that at open. Updating the last reference
date at close ensures that data sets are not automatically migrated
based on the last reference date at open in the DFSMShsm space
management cycle after a subsystem is brought down after many days.
− Data class constructs
All VSAM data set attributes are available from the data class using
Interactive Storage Management Facility (ISMF). The enhancements to
data class for VSAM support the BWO, LOG, LOGSTREAMID, SPANNED,
and NONSPANNED attributes. Data sets defined with JCL can use these
attributes if they are assigned a data class having these attributes.
− VSAM record level sharing (RLS) extended addressability
This enhancement allows RLS access to VSAM KSDSs defined with
extended addressability beyond 4GB.
− DCOLLECT enhancements
DCOLLECT is enhanced to collect more information as well as to have
more granularity to obtain information about specific volumes.
• DFSMShsm
− Many enhancements to ABARS have been made in this release.
− Access Method Services for ICF (IDCAMS) allows altering the storage
and management class without recalling a migrated data set.
− DFSMShsm provides an option for its control data sets (CDSs) to be
accessed in RLS mode. The ability to use VSAM RLS to access the
DFSMShsm CDSs will eliminate data set contention against the CDSs
when three or more hosts are processing in a shared environment.
− DFSMShsm uses dump analysis elimination (DAE) to suppress duplicate
dumps.
− DFSMShsm provides an alternative to TAPECOPY processing for backup
and migration cartridge tapes. This option allows DFSMShsm to create
two tapes concurrently. It is intended that the original tape be kept
onsite and the alternate taken offsite or written to a remote tape library.
− DFSMSrmm no longer uses the DFSMShsm installationwide exit,
ARCTVEXT. It provides a new general-use programming interface,
EDGTVEXT, for releasing tape volumes. Thus ARCTVEXT is freed from
DFSMSrmm usage, allowing it to be used by other tape management
systems.
− Many serviceability items have been added to this release of
DFSMShsm.

Chapter 1. Introduction to DFSMS/MVS Version 1 Release 4 3

 • DFSMSdss
− DFSMSdss Multivolume Selection and Filtering
A new data set selection keyword, SELECTMULTI, acts in conjunction with
the user-supplied input volume list to control the selection of data sets to
be copied, dumped, or converted to or from SMS management.
• DFSMSrmm
− New parameters have been added to the EDGRMMxx member of
PARMLIB that allow an installation to specify a percentage-full threshold
and the name of an automatic backup procedure.
− DFSMSrmm will no longer provide its own version of the ARCTVEXT exit.
− The Problem Determination Aid (PDA) trace is added to DFSMSrmm so
that diagnostic information can be gathered at entry and exit points in
the DFSMSrmm modules.
− DFSMSrmm now allows an installation to use concurrent copy to back up
the control data set and journal.
− Improved support is provided for DFSMShsm alternate tape processing.
− DFSMSrmm can now be instructed to manage a tape volume but not to
track all of the data sets that are written to that volume. With this
enhancement DFSMSrmm and other tape management systems will no
longer maintain duplicate information.
− DFSMSrmm provides a trial-run capability for inventory management.
Thus you can check what DFSMSrmm will do during VRSEL processing
without making changes to the control data set or journal.
• Open and distributed enhancements
− Network File System (NFS) client support
A new NFS client is provided by the DFSMS/MVS V1R3 NFS Feature for
the OS/390 and MVS platforms. Thus one OS/390 (or MVS) system can
exchange data files with another, using the NFS client/server technology.
The DFSMS/MVS NFS Client provides the full capability for OpenEdition
applications to read or write MVS conventional data sets and
OpenEdition Hierarchical File System (HFS) files on another OS/390
system.
− Distributed File Manager (DFM) DataAgent
The DFM DataAgent is an extension to the DFM and the VSAM/x
component of SmartData Utilities (SdU) on Windows, OS/2, and RS/6000.
The DFM DataAgent allows users of SdU to issue TSO commands and
access otherwise unsupported MVS data sets or databases. The DFM
DataAgent provides flexibility in allowing client applications access to the
data itself rather than limiting access to predefined remote transactions.

1.1.2 Usability Enhancements in DFSMS/MVS V1R4

• NaviQuest
The DFSMSdfp NaviQuest component enables you to perform SMS migration
functions and many routine storage management tasks, including testing and
reporting, in batch.
• Catalog Search Interface (CSI)

4 DFSMS/MVS V1R4 Technical Guide

 CSI provides an application programming interface (API) from assembler and
high-level languages to read information from Integrated Catalog Facility
(ICF) catalogs. It is an alternative to LISTCAT that allows tailoring of the
output and includes some information that LISTCAT does not provide.
• DFSMS Optimizer
The DFSMS Optimizer product has been totally refreshed. Even though it is
not a DFSMS/MVS V1R4 line item, an overview of the product and its function
is included in this book. See Chapter 8, “DFSMS/MVS Optimizer” on
page 115.

Chapter 1. Introduction to DFSMS/MVS Version 1 Release 4 5

6 DFSMS/MVS V1R4 Technical Guide
Chapter 2. DFSMSdfp

The DFSMSdfp component in DFSMS/MVS V1R4 provides usability and

performance enhancements, greater reporting capabilities, and enhanced
protection of checkpointed data sets and satisfies many customer requirements.
In this chapter we cover the following DFSMSdfp enhancements:
• Space allocation failure reduction
• SAM tailored compression
• O/C/EOV serviceability enhancements
• Enhanced protection of checkpointed data sets
• Optical Access Method SMF recording
• Program Management 3 (PM3)

2.1 Space Allocation Failure Reduction

Sometimes a data set allocation or extension fails because there is not enough
space on a volume to satisfy the space allocation request using current
algorithms. DFSMS/MVS solves this problem to some extent by performing
volume selection and trying all candidate volumes before failing an allocation .

The space allocation failure reduction further reduces the incidence of these
allocation and extension failures. It applies to SMS-managed data sets only and
is limited to new data set allocations and extends on new volumes. The space
allocation failure reduction enhancement does not provide relief for data set
extensions on the current volume.

The following enhancements may make having separate storage groups for
large, medium, and small data sets less of a requirement:
• The five-extent limit in DADSM for new data set allocation and for extends to
new volumes has been removed.
• Data sets can grow to either 59 volumes (which is the limit imposed for
multivolume data sets by the task input/output table (TIOT)) or the number of
volumes in the storage group (as long as a volume in the storage group has
space), whichever is smaller.
• Maximum number of extents per component of VSAM data sets is now 255
instead of 123 for multivolume data sets.
• Data sets that currently have a 16-extent or 123-extent per volume limit will
continue to have the same limit.

Solving the allocation and extent problem results in the added benefit of not
having to run DEFRAG as often, especially in a sysplex environment. DEFRAG
can move extents of a data set only when it is not in use anywhere in the
sysplex and requires GRS ring/star or equivalent product support to ensure data
integrity.

 Copyright IBM Corp. 1997 7

2.1.1 Allocations Affected
The allocations affected by the space allocation failure reduction are:
• Allocations initiated by BATCH (JCL) or dynamic allocation (SVC 99)
• Allocations initiated by IDCAMS DEFINE (VSAM only) and IDCAMS or TSO
ALLOCATE commands
• Allocations initiated by DFSMSdss or DFSMShsm.
DFSMSdss and DFSMShsm only use the enhancement that removes the
five-extent limit. The space allocated will not be reduced.
• Data set extensions to new volumes
Note that the abend x37 reduction does not address data set extensions on
the current volume.
• VSAM allocations
VSAM allocations are affected by an increase in the maximum number of
extents per VSAM component from 123 to 255. The maximum number of
extents per data set per volume remains at 123.
Note: DFSMSdss or DFSMShsm initiates allocations to allocate space for data
sets that must be restored, and the allocations cannot tolerate a reduction in the
requested space amount. Therefore, the %X value in the data class is ignored
for these types of allocations.

2.1.2 Implementation Details

As data set allocations proceed, one of the following may occur:
• The allocation is successful.
• The allocation fails because enough space was not available.
• The allocation fails for a reason other than space.

Of the two failing cases (case 2 and case 3), space outage reduction addresses
case 2. It is based on retrying allocation failures by a combination of:
• Spreading the requested space quantity over multiple volumes
• Reducing the requested space quantity by the allowed limit
• Using more than five extents to satisfy the allocation

If a volume selection failure occurs for a reason other than space, the space
outage reduction solution will not apply.
Note: Spreading data over multiple volumes means that SMS allocates the
primary space allocation over as few volumes as possible under the constraints
imposed by the maximum number of extents that may be allocated on each
volume. (For VSAM data sets, the constraints are based on the number of
extents allowed for the entire component over all volumes.)

2.1.3 Volume Selection Failure Recovery

Users may, through use of newly defined fields in the data class , indicate to SMS
that they wish to attempt a retry using one or more of the above methods.
These new fields are:
• SPACE CONSTRAINT RELIEF: Y|N (default is N)
If a user specifies Y for space constraint relief:

8 DFSMS/MVS V1R4 Technical Guide

 − On initial allocations, more than one volume will be used to allocate the
primary quantity if the data set is multivolume eligible.
− SMS may reduce the amount of space to be allocated on the basis of the
REDUCE SPACE UP TO parameter.
− SMS may remove the five-extent limit.
• REDUCE SPACE UP TO (%): X(0 to 99, 0 is the default)

In the paragraphs that follow, we illustrate how SMS processes these new fields.

2.1.4 New Allocations (VSAM and non-VSAM)

If a user specifies N for SPACE CONSTRAINT RELIEF, either explicitly or
implicitly, SMS will not attempt a retry. Allocation will not be attempted by
spreading the data on multiple volumes.

Retry is to allocate the reduced space quantity in as many extents as allowed by

the removal of the DADSM five-extent limit.

If a user specifies Y for SPACE CONSTRAINT RELIEF and the volume count for
the allocation is 1, SMS will redrive the allocation after reducing the requested
space quantity (which could be PRIMARY or SECONDARY space quantity) on the
basis of the REDUCE SPACE UP TO parameter. If the reduced space quantity
(the requested space multiplied by the REDUCE SPACE UP TO parameter) is less
than 1, it will be rounded up to 1. Simultaneously SMS will also allow more than
five extents to allocate this recomputed space amount.
Note: It is valid for a user to specify 0 space quantity for primary space (when
allocating a model data set control block). The space quantity for non-VSAM data
sets is the PRIMARY space for new allocations and the SECONDARY space for
extensions. The space quantity for EF and non-EF VSAM data sets is the
PRIMARY space for new allocations and either the PRIMARY or the SECONDARY
space for extensions on new volumes (depending on a parameter in the data
class). Such allocations should not fail for space except in the rare
circumstance where there is no space in the VTOC for allocating a DSCB. In any
event, allocations where the primary space quantity is 0 will not be retried.

If the user specifies a Y and the volume count for the allocation is more than 1,
on retry SMS will attempt to allocate the requested space using more than one
but as few volumes as possible.

For allocation, the maximum number of volumes is equal to the largest of the
unit count, volume count, and volser count. If none of these three values is
specified, the volume count in the data class definition is used. This volume
selection failure recovery method is called best-fit allocation . It is applicable
only during the initial allocation of a data set. It is not applicable during
extensions to new volumes.
Note: For non-VSAM data sets, if the secondary space quantity is zero, the data
set is not extended on the current or new volume.

The amount of space allocated during extent processing of VSAM data sets
differs from that of non-VSAM data sets. If a VSAM data set is extended on the
current volume, the SECONDARY space specified by the user is allocated. If the
VSAM data set is extended to a new volume, the PRIMARY or SECONDARY
space specified is allocated as follows: the default is PRIMARY space; the

Chapter 2. DFSMSdfp 9
 SECONDARY quantity is used if the data class allows it and the data set is
allocated in extended format.

If the best-fit allocation fails, SMS will retry the allocation as follows:
• If REDUCE SPACE UP TO (X%) is specified and X is between 0 and 99, SMS
will reduce the requested space quantity by X% and redrive the best-fit
allocation. A specification of 0 implies that the user wants to use more than
five extents to satisfy the allocation without reducing the allocation amount.
• SMS will use as many extents as are allowed.

2.1.5 Extents on New Volumes

Extensions will be tried only if candidate volumes exist for the data set as
indicated in the TIOT.

2.1.6 Non-VSAM Data Sets

If the initial attempt to extend the data set fails and the data class indicates that
SPACE CONSTRAINT RELIEF is applied, SMS will retry the extents as follows:
1. If the user has specified (with the REDUCE SPACE UP TO ) parameter that a
smaller amount of space is acceptable, SMS will reduce the amount of space
and will indicate to DADSM that the five-extent limit is to be removed.
2. If the user has indicated that space is not to be reduced, by either specifying
0 for REDUCE SPACE UP TO or defaulting to 0, SMS will indicate to DADSM
that only the five-extent limit is to be removed.

2.1.7 VSAM Data Sets

If the initial attempt to extend the data set fails and the data class indicates that
SPACE CONSTRAINT RELIEF is to be applied, SMS will retry the extent using the
smaller of up to 123 extents per volume or 255 extents for the VSAM component.

2.1.8 Exceptions to Using Volume Selection Failure Recovery

The following types of data set allocations cannot use volume selection failure
recovery:
• All allocations where the storage class specifies a nonzero sustained data
rate (SDR), implying a multistriped data set, and the data class specifies
REQUIRED for striping. With this exception, all VSAM extended format data
sets (which are always single striped) will be candidates for space constraint
relief. Furthermore, all non-VSAM extended format data sets that specify a 0
SDR (which results in single striping) will be candidates for space constraint
relief. All non-VSAM extended format data sets with SDR other than 0 will
not be candidates for space constraint relief.
• All multivolume guaranteed space allocations (VSAM and non-VSAM) will be
retried with the five-extent limit removed, but the space quantity will not be
reduced.
• VSAM key-range data sets will not be candidates for space constraint relief.
• If IMBED is specified (VSAM data sets only), SMS will not use the best-fit
allocation method during the redrive because it is not supported in
conjunction with IMBED. SMS will still attempt the redrive with greater than
five extents and a smaller space quantity (if specified by the user).

10 DFSMS/MVS V1R4 Technical Guide

2.1.9 ISMF Changes
The data class Application Selection panels will be enhanced for the following
options:
• Data class Define/Alter
• Data class Display
• Data class List Display
• Data class List Sort
• Data class List View
• Data class List Print

The data class Define/Alter page 3 panel, DGTDCDC5, and the data class Display
page 3 panel, DGTICDC3, have been modified to accommodate the new SPACE
CONSTRAINT RELIEF and REDUCE SPACE UP TO fields. The fields have been
added to the last page, at the end, after all existing fields.

Data class List Display panel DGTLGP21, List Sort page 2 panel DGTDCDC4, List
View page 2 panel DGTDVW42, and List Print page 2 panel DGTDPR42 have
been modified to accommodate the SPACE CONSTRAINT RELIEF and REDUCE
SPACE UP TO new columns. They are to be added to the end or existing entries
are to be moved down or across columns to accommodate the new entries in
the appropriate place.

2.1.10 ISMF Message Changes

ISMF has been enhanced with the following new messages on the data class
Define/Alter panels:

DGTDC060 INVALID VALUE

SPACE CONSTRAINT RELIEF must be Y or N

DGTDC061 INVALID VALUE

REDUCE SPACE UP TO must be 0 to 99 when SPACE CONSTRAINT RELIEF is Y

DGTDC062 FIELD RESET TO DEFAULT

REDUCE SPACE UP TO is reset to 0, since SPACE CONSTRAINT RELIEF is Y

DGTDC063 FIELD RESET TO DEFAULT

REDUCE SPACE UP TO is reset to Blank, since SPACE CONSTRAINT RELIEF is N

2.1.11 Migration and Coexistence

The following areas might be impacted by the space outage reduction changes:
• Storage administrators
Storage administrators may not have to schedule DEFRAG as often, and they
may not have to have separate storage groups based on data set size to
avoid space failures.
Storage administrators must determine whether or not to modify existing
data classes and/or define new data classes with the new SPACE
CONSTRAINT RELIEF and REDUCE SPACE UP TO attributes. Generally

Chapter 2. DFSMSdfp 11
 speaking, it is beneficial to specify Y for SPACE CONSTRAINT RELIEF and a
nonzero value for REDUCE SPACE UP TO unless the user application cannot
tolerate a reduction in the amount of space requested. Data class changes
apply to both new and existing data sets.
• Application programmer
Application programmers must be aware of the new messages that they may
encounter. If they expect the requested space to be satisfied within five
extents and on a single volume, the changes introduced by space outage
reduction could affect applications. If the new data class attribute
parameters are not used, applications will not be affected.

PTFs will be shipped on supported releases of MVS/DFP and DFSMS/MVS to

support 255 extents per VSAM component.

With these PTFs applied, lower-level systems can access all 255 extents of a
VSAM component that was created on a DFSMS/MVS V1R4 system. It will not
be possible, however, to extend VSAM components on lower-level systems if
they occupy 123 or more extents.

If customers do not want to use the new parameters in the data class, they need
take no action, and the system will continue to work as before. If customers
want to use the new parameters, they have to define new data classes or modify
existing definitions.

Customers specifying SPACE CONSTRAINT RELIEF in one or more data classes

should note the following points:
• Very large allocations that would have failed may succeed if a sufficiently
large volume count is specified in the data class or JCL.
• Existing data sets may end up with less space than requested on extents just
by changing a data class and activating a configuration.
• The space allocated for new data sets may be less than requested, again
depending on the use of new attributes in the data class.
• The number of extents used during initial allocation may result in fewer
extensions to new volumes or current volumes.
• There should be fewer X37 abends.

2.2 SAM Tailored Compression

This enhancement significantly improves the compression ratios that can be
achieved for sequential data sets accessed by BSAM or QSAM. Tailored
compression introduces a new form of compression for sequential extended
format data sets. With tailored compression, the system attempts to derive a
compression dictionary that is tailored specifically to the initial data written to a
data set. Once a tailored dictionary is derived, it is imbedded in the compressed
data set. This technique is expected to provide improved compression ratios,
thereby reducing DASD usage and channel traffic. Because the dictionary is
tailored to the user data, significantly more data is sampled than was required
by generic compression. The process of sampling the data and building the
dictionary during creation of a new data set takes more CPU cycles and is,
therefore, most noticeable when compressing small data sets. For larger data
sets, the cost of sampling is amortized significantly. Reuse of a tailored
compressed data set saves the cost of sampling and dictionary creation.

12 DFSMS/MVS V1R4 Technical Guide

 An installation has the option of either using the new tailored compression or
continuing to use generic DBB-based compression first introduced with
DFSMS/MVS V1R2.

Tailored compression allows for more types of data to be compressed, for

example, where non-English languages are used. Because the original generic
dictionary was developed with standard American-English scripts, files
containing sequences of characters that do not appear in the American scripts
do not compress very well. The tailored dictionary avoids this problem as it is
generated dynamically from the data itself, for each data set. Dynamically
generating a tailored dictionary from the data itself should make tailored
compression more useful to the non-English-speaking world.

Tailored compression support does not apply to VSAM KSDSs, which can
continue to be compressed with generic DBB dictionary compression (as
illustrated in Figure 1).

Figure 1. Tailored Compression. The IGDSMxx PARMLIB m e m b e r specifies whether tailored compression or
generic DBB compression will be used. VSAM KSDSs can only use generic compression.

2.2.1 Tailored Compression and Generic Compression

The existing generic compression algorithm was used for both VSAM KSDSs and
SAM data sets. Standard DBBs continue to be shipped with the product in a
system library called SYS1.DBBLIB and will be used whenever:
• A VSAM data set is compressed.
• The installation specifies generic as the default compression algorithm.
• A data set was previously compressed using the GENERIC option or was
compressed on a pre-DFSMS/MVS V1R4 system.

Chapter 2. DFSMSdfp 13
 Standard DBBs consist of a fixed set of padding run blocks, numeric blocks, and
English text strings commonly found in the English language. As such, standard
generic compression algorithms do not work well for non-English languages, or
any text data that involves repetitive strings other than common English words;
for example, any data set made up of information not found in the available
DBBs, such as the repeating occurrence of a person′s name.

In contrast to generic, tailored dictionaries are built from data strings extracted
from the data to be compressed, assembled into a compression dictionary, and
stored in the first few tracks of a SAM data set; that is, tailored compression is
completely insensitive to the text language. As with generic, tailored
compression samples the first portion of data as it is written, choosing the
optimal data strings to use. These data strings are then assembled into a
compression dictionary and stored into the data set. This compression
dictionary is read and activated whenever the data set is opened.

Tailored compression′s design requires it to sample significantly more data than

was required by the generic algorithms (400K on average verses 64K). This
additional sampling uses more CPU cycles and is therefore most noticeable
when compressing small data sets. Because the cost of tailored compression
sampling is significantly higher than generic sampling, generic is the default
system option.

2.2.2 Sampling the Data

For larger data sets, the cost of sampling is amortized significantly. If a small
data set is opened for output a second time, the sampling process is not
repeated. During subsequent opens of a data set, the already created
compression dictionary is read from disk and is reused. Reuse of the
compression dictionary saves CPU time, but it also means that if the nature of
the new data is very different, it is essential that the data set be scratched and
reallocated in order to force the system to re-create the compression dictionary.

Evidence shows that for large data sets, tailored compression significantly
improves the data compression ratio while it reduces the CPU time needed to
compress a data set (when compared to generic), although this CPU time is still
much higher than not using compression at all. Therefore, customers must still
make a trade-off between CPU time costs, channel traffic, and DASD space
savings, but tailored compression significantly reduces DASD space usage for
small as well as large data sets, even when taking into consideration the cost of
storing the compression dictionary. Also, because less DASD space is used, the
elapsed time to read or write the data set is reduced.

When evaluating whether or not to use compression for all data sets, it is
important to take into consideration factors other than the size of the data sets,
such as how long the data set is expected to be kept. When a data set is reused
(either read or rewritten), the CPU cost is much less. Therefore, it may be
desirable to restructure applications so that sampling takes place less
frequently.

Compression is never used for temporary data sets. The above discussion
applies only to permanent data sets.

14 DFSMS/MVS V1R4 Technical Guide

2.2.3 Enablement of Tailored Compression
A new parameter, COMPRESS(TAILORED|GENERIC) is available for the IGDSMSxx
member of SYS1.PARMLIB. It enables you to specify the form of compression
the system is to use for newly created physical sequential extended format data
sets.

COMPRESS(GENERIC) refers to DFSMS/MVS V1R2 generic DBB-based compression.

This is the default parameter to maintain consistency with previous DFSMS/MVS
function. However, IBM generally recommends that customers change the
option to tailored compression. When all the systems in the sysplex and in the
disaster recovery site are not at the DFSMS/MVS V1R4 level, specify
COMPRESS(GENERIC).

COMPRESS(TAILORED) refers to tailored compression where the system attempts to

derive a dictionary tailored specifically to the initial data written to the data set.

Once the COMPRESS(TAILORED) parameter is activated (through the SETSMS=xx

command or an IPL), the system uses tailored compression for newly created
sequential compressed format data sets.

Customers can choose at the data set level through a DATACLAS parameter
whether or not a data set should be compressed.

2.2.4 Wellness Logic

A new feature of DFSMS/MVS V1R4 compression is called wellness logic (see
Figure 2). If DFSMS recognizes that the data set was not compressing to a
desired level, it stops the compression process for a set period of calls. After
this skip duration, data compression resumes, and the wellness logic is again
activated. If a data set consistently compresses poorly, over the entire length of
the data set DFSMS samples 20% of the time and 20% of the data is
compressed.

For example, suppose that in DFSMS/MVS V1R3 a data set was compressed by
only 8% and the CPU time cost was 1 sec. In DFSMS/MVS V1R4, the data set
would be compressed by 1.6%, but the CPU time for sampling would be reduced
to 0.2 sec (as illustrated in Figure 2).

-----DS Size (MB)------- Compression CPU seconds

---GENERIC-- TCOM --GENERIC--- TCOM
Data Non- DFSNS DFSNS DFSMS SFSMS
Set Comp. 1.3.0 1.4.0 1.3.0 1.4.0 DISP=NEW
6 9.4 8.6 9.2 3.3 3.8 0.8 3.1

Figure 2. Wellness Logic

Wellness logic applies to both generic and tailored compression. However,

tailored compression almost always exceeds the wellness criteria, so wellness
logic will very rarely come into play for tailored compression.

Chapter 2. DFSMSdfp 15
2.2.5 Measurements
All measurements were done using a 9672-RX73 processor and IEBGENER to
copy an existing data set. The CPU cost of compression was derived by first
measuring the CPU time when only the output data set is compressed and then
subtracting the CPU time when both the input and output data sets are
noncompressed. Likewise, the CPU cost of decompression was derived by first
measuring the CPU time when only the input data set is compressed and then
subtracting the CPU time when both the input and output data sets are
noncompressed.

In the case of tailored compression, two variations were measured. The first
time IEBGENER was run is called DISP=NEW, which includes the cost of
sampling. The second time IEBGENER was run is called DISP=OLD, which
avoids the cost of sampling. Generic compression was not measured with
DISP=OLD, but it is expected that the cost of generic sampling is relatively
small compared to the cost of tailored compression sampling.

In Figure 3, data sets 2 through 5, generic compression achieved a compression

ratio around 1.8. In contrast, for these same data sets, tailored compression
achieved a compression ratio between 2.7 and 4.2. For data set 1 where generic
compression achieved only a 13% size reduction, tailored compression achieved
a compression ratio of 2.7.

---DS Size (MB)---- -------Compression------- Decompression

Data Non- -------CPU seconds------- CPU seconds
Set Comp. GENERIC TCOM GENERIC ------ TCOM ----- GENERIC TCOM
DISP=NEW DISP=OLD
1 420 367 155 177.0 148.6 143.1 14.6 12.8
2 28.5 15.5 6.7 10.6 16.1 9.6 0.9 0.6
3 16.1 8.7 4.9 6.4 13.3 5.7 0.5 0.6
4 13.5 7.3 4.9 5.2 12.7 4.8 0.5 0.4
5 6.4 3.8 1.9 3.1 7.2 2.2 0.4 0.2

Figure 3. Generic and Tailored Compression Measurements

Note that although most of these data sets might be considered small or medium
size data sets, the same compression ratios would occur if the data were
replicated to create larger data sets.

For tailored compression with DISP=OLD, the normalized CPU time per kilobyte
for each of these data sets is in a narrow range between 330 and 355 sec/KB.
(This range is sensitive to the CPU model.) Although generic DISP=OLD
measurements were not done, it appears likely that such a metric would land in
the same range as with tailored compression, except for data set 1, which
generic compression does not compress well. In those cases where data does
not compress well, the CPU time per KB is higher.

Data set 6 in Figure 2 on page 15 demonstrates the wellness logic. Generic

compression achieved less than 10% size reduction. In DFSMS/MVS V1R3
where wellness logic did not exist, DFSMS still tried to compress the data set
and the CPU time cost was very high. In DFSMS/MVS V1R4, the wellness logic
with generic compression reduced the CPU cost significantly, with only a slight
change in the data set size. On the other hand, tailored compression
compressed the data set significantly, with a corresponding cost in CPU time.

16 DFSMS/MVS V1R4 Technical Guide

2.2.6 Functional Characteristics
All sequential data sets that are eligible are compressed with the compression
type specified in the IGDSMxx member. This is true even if the data set is
allocated with LIKE and modeled after a data set with a different compression
type.

If it is determined that the data set is eligible for tailored compression, the new
function:
• Builds a tailored dictionary using the initial data written to the data set
• Validates the useability of the resulting dictionary depending on whether a
preset compression criterion is met.
• Compresses the data set (including the initial sampled data) or rejects the
compression for the data set if a preset compression criterion is not met.

2.2.6.1 Tailored Compression Identification

Note: The following information is not an intended programming interface. It is
provided for diagnostic purposes only.

A dictionary token consists of 32 bytes and is stored in the catalog. Today, the
dictionary token may be found in output displayed by functions such as LISTCAT
and DCOLLECT, and in certain SMF records. The dictionary token is modified to
contain tailored compressed data set indicators. The first byte of the dictionary
token indicates the type of compression used for the data set (see Figure 4 on
page 18):

• x′ 1 0 0 . . . . . ′ indicates compression has been rejected for the data set,

no data is compressed
• x′ 0 1 0 . . . . . ′ indicates generic DBB compression is used
• x′ 0 1 1 . . . . . ′ indicates tailored compression is used

The second four bits in the first byte of the token contain bits to denote the
number of blocks containing system data from the beginning of the data set until
the user data starts:

• x′ . . . . . xxx′ indicates the number of blocks containing system data

The start of user data will begin in the physical block following the blocks
containing system data. A compressed data set with a tailored token or a
rejection token may contain a number of these system data blocks.

Chapter 2. DFSMSdfp 17
Figure 4. Dictionary Token. Tailored compression is indicated from the first four bits of the first byte.

For tailored compressed data sets, the number of system blocks is always
nonzero, as this is where the tailored dictionary is stored. When a compressed
format data set contains a rejection token, the number of system blocks may still
be nonzero.

When a compressed data set has system blocks, these blocks are reflected in
the DS1LSTAR field in the data set′s Format 1 DSCB. On reusing an existing
data set, it is possible for a compressed data set to contain no user data but still
have a nonzero DS1LSTAR.

There may be a significant increase in the amount of data DFSMS compression

buffers in storage for new data sets while it derives the dictionary. This increase
allows DFSMS compression to optimize dictionary building and compress
sampled data. This compares to using the generic DBB-based dictionary where
up to the first 64KB of user data may be stored in uncompressed form.

2.2.6.2 Storage Requirements

Because new internal compression control blocks are being used, together with
the increased buffering of data during the dictionary build phase, a minimum
region size of 3MB is required. This storage will be getmained from above
16MB. Once a dictionary is derived, the ongoing storage requirements are
reduced to less than 200KB.

The REGION size parameters of the JOB and EXEC JCL statements do not have
to be modified to use tailored compression, even though 3MB more storage
above 16MB is used.

2.2.7 Migration and Coexistence Considerations

A data set compressed with a tailored dictionary cannot be processed on any
previous to DFSMS/MVS V1R4 system. Compatibility PTFs are required on
pre-DFSMS/MVS V1R4 systems to fail attempts to access these data sets on
those systems. On a system with tailored compression active, it will not be
possible to convert, in place, existing data sets with or without generic

18 DFSMS/MVS V1R4 Technical Guide

 dictionaries to a tailored dictionary data set. You must copy the data into
another data set, using a utility like IEBGENER or an IDCAMS REPRO.

Any attempt to access a data set using a tailored dictionary on a

pre-DFSMS/MVS V1R4 system through DFSMSdss or DFSMShsm will fail. Error
message ADR910E will be issued to describe the reason for the failure.

Tailored compression does not apply to VSAM data sets at this time. Even
though the system option might specify tailored compression, VSAM will use
only standard generic dictionaries.

2.3 O/C/EOV Serviceability Enhancements

Customers, service, and development personnel have documented a number of
O/C/EOV serviceability suggestions. DFSMSdfp V1R4 will be enhanced to satisfy
several of these serviceability suggestions, such as more information when an
x37 abend occurs by providing a new SMF subtype record. A trace will be also
available as well as a new message in cases of O/C/EOV problems.

O/C/EOV provides serviceability enhancements to help customers diagnose the

problems by applying information in the SMF record 42 and trace and format the
O/C/EOV work area.

The following DFSMSdfp enhancements are introduced in DFSMS/MVS V1R4:

• The O/C/EOV problem determination optional work area trace provides
module flow and trace information for DFSMSrmm tape exit modules as is
now provided for all other O/C/EOV modules. Thus IBM as well as customer
services personnel can determine whether or not the DFSMSrmm tape exit
modules were executed.
• A LOGREC is written for Automated Tape Library (ATL) devices whenever an
abend 613 (an error occurred during processing of an OPEN macro
instruction for a data set on magnetic tape) or abend 637 (an error occurred
at an end-of-volume for a data set on magnetic tape or an end-of-volume
during concatenation) is issued. This record will provide useful first-time
data capture information that could help resolve the cause of the abnormal
task termination. The information could include the unit control block (UCB),
UCB extension, LACS parameter list, and other pertinent ATL data (such as
the token and external volser). The message text for IEC214I and IEC026I will
be updated to document the LOGREC identifier.
• A new started task enables the O/C/EOV problem determination optional
work area trace to be run easily for dynamically allocated data sets. This
task is started by the operator and then requests information such as the
data set name, job name, and the DD statement name of the data set to be
traced. The task can also be used for batch jobs, thereby eliminating the
need to add DCB=DIAGNS=TRACE to the DD statement of the data set to
be traced. The maximum number of trace entries is 10. Other functions
provided are the ability to display the active entries and to delete either one
or all of them. O/C/EOV checks the trace table to determine whether the
data set currently being processed should be traced just as if
DCB=DIAGNS=TRACE were specified on the DD statement. New
messages, IEC980A and IEC980I, are used to interface with and provide
information to the operator.

Chapter 2. DFSMSdfp 19
 • The O/C/EOV abend message reflects the name of the module that detected
the error instead of the name of the module that issued the abend. For
example, during a B37, D37, or E37 abend, module IFG0554T is documented
in the message text, even though module IFG0554P actually requested the
abend. The message text now indicates IFG0554P.
• Currently if a user tries to open a PDS for output by specifying DISP=SHR,
and the PDS is already open in this condition, the user has failed to serialize
access before attempting to open the data set. The user has no information
about who owns the PDS for output. The data set might be in the same
system or another system that is sharing the volume. O/C/EOV is enhanced
to report helpful information to find out who has the PDS for output. If an
abend 213 reason code 30 is issued, a new message, IEC813I, is issued
documenting which address space, JOB, and task control block (TCB) owns
the PDS resource that is preventing the open from being allowed. The
information provided by the IEC813I message provides first-time data capture
information.
• A new SMF type 42 subtype 9 record is written for B37, D37, and E37 abends,
documenting such information as job name, data set name, volser, number
of extents on the volume, and secondary allocation amount that system
programmers can use to prevent x37 abends in the future.
• An above 16MB O/C/EOV work area extension has been implemented. This
extension does not provide new function; it has been implemented because
of storage constraints in the main O/C/EOV work area.

2.3.1 New Trace Processing

Trace setup processing now runs as a started task. It is started, ended, and
modified by operators.

The new started task module, IFG0OCET, is installed in SYS1.LINKLIB. The

started task procedure, IFGOCETR, is installed in SYS1.PROCLIB.

2.3.2 Enabling IFGOCETR

The steps listed below must be followed to activate the O/C/EOV problem
determination work area trace started task for dynamically allocated data sets:
1. The application or system programmer must supply the operator with either
the data set name, job name, or DD name (or any combination thereof) of the
data sets to be traced. As much information as required to uniquely identify
the data set to be traced should be supplied. For instance, if only a job
name is specified, then all data sets in the jobs matching that job name will
be traced. This is probably not what the application or system programmer
intended.
2. The operator starts the task by entering:
• START IFGOCETR
3. IFGOCET issues WTOR message IEC980A, requesting the data set name, job
name and/or DD name for the data set to be traced. There can be up to 10
entries in the trace table. O/C/EOV checks the active trace table entries to
determine whether the data set being processed should be traced. After a
trace entry has been added to the table, the active trace entries are
automatically displayed. When an entry is added to the table, all data set
parameters (data set name, DD name, and/or job name) must be specified
on one response to message IEC980A. In other words, if the data set name

20 DFSMS/MVS V1R4 Technical Guide

 is specified in one response and the DD name is specified in the next
response, two separate table entries will result. The data set name and DD
name must be specified in one response to create only one table entry. The
DISPLAY and DELETE parameters can be specified at any time but not in the
same response as the data set name, DD name, or job name. When all
responses have been issued, the operator must respond “END” to the
messages to automatically end the IFGOCETR task. IFGOCETR does not
have to be executing in order for data sets to be traced. Once the table has
been updated, IFGOCETR can be terminated.
4. The operator starts the generalized trace facility (GTF) as usual (specifying
USR=(FFF)). No new GTF parameters are required to use this new function
and, once the trace data has been obtained, the GTF started task must be
stopped as usual.
5. The operator can display trace entries and delete either one or all entries
from the trace table.
6. Message IEC980I is issued for any error or informational conditions.

2.3.3 IFGOCETR Parameters

The IFGOCETR parameters are supplied by the operator in responding to
message IEC980A. The parameters are:
• DSN= d s n (dsn is the data set name to be traced)
• DDN= d d n a m e (ddname is the DD name of the data set to be traced)
• JN= j o b n a m e (jobname is the job name to be traced)
• DISPLAY (displays all active trace entries)
• DELETE,<ALL|entno>
− All deletes all active entries
− Active trace entry entno is disabled
• END (All trace parameters have been entered. This causes the started task
to terminate.)
Note: At a minimum, either DSN, DDN, or JN must be specified, but any
combination of those parameters can be specified. In fact, we recommend
specifying a combination to limit the number of data sets that will be traced. For
example, if only JN is specified, all data sets processed by the job will be traced.
(This is probably not what was intended; however, it will work.) So the
combination of DSN, DDN, and JN should be used to limit the tracing to one data
set.

Parameters DISPLAY and DELETE are mutually exclusive from DSN, DDN, and
JN. If DISPLAY or DELETE is specified with DSN, DDN, or JN, it will be ignored.
DSN, DDN, and JN can be specified only once in each operator response. If
multiple trace entries are to be created, each must be created separately.

If the data set to be traced is a generation data group (GDG), DSN can be either
the complete generation data set (GDS) name (including G000v00) or the GDG
base name. If the complete GDG name is specified, only that generation data
set will be traced. If the GDG base name is specified, all of the generations are
eligible for tracing, depending on the other parameters specified.

Chapter 2. DFSMSdfp 21
 The started task does not have to be active at the time of the data set tracing.
The started task is used only so the operator can enter parameters and cause
the trace table to be built or updated. The trace table, once obtained, is not
specifically freed. It will be allocated and available for use during the entire
duration of IPL. Only by re-IPLing is the table freed.

If the trace table becomes full, trace table entries must be deleted before a new
entry can be added.

2.3.4 Migration and Coexistence

This O/C/EOV serviceability enhancements introduce no specific migration
considerations for customers. However, OEM tape vendors may be interested in
the new work area extension. They should be encouraged to migrate away from
hooks in O/C/EOV modules to using the tape installation exits that are now
provided. A good reference for details on tape exits is DFSMS/MVS Installation
Exits (SC26-4908).

2.4 Enhanced Protection of Checkpointed Sequential DASD Data Sets

If a customer uses both MVS and IMS GSAM checkpointed data sets, it is not
possible to distinguish them. That is, DFSMSdfp can recognize whether or not
there is a checkpointed data set, but it cannot tell whether it is an MVS
checkpointed (MVSC) or an IMS GSAM (IMSC) checkpointed data set. It is
important to be able to distinguish these data sets because of the way by which
MVS and IMS access them. MVS accesses checkpointed data sets by absolute
addressing (cylinder, head, record), so the checkpointed data set must not be
movable. If the checkpointed data set is moved, restarts will fail. IMS accesses
checkpointed data sets by offset addressing (from the beginning of the data set
on the volume). Therefore, the data set is movable if the device characteristics
and the amount of data per volume does not change.

In MVS/DFP V3R2 an indicator was provided in the format 1 DSCB called

DS1CPOIT , which would be set in the MVS checkpoint supervisor call instruction
(SVC) for physical sequential data sets and in DFP Open for IMS GSAM data
sets. Notice that as far as DFP is concerned, GSAM data sets are physical
sequential data sets. Customers were expected to use the DFSMShsm
automatic migration exit to disallow migration until the data set was no longer
required for a restart. They were expected to use the creation date, last
reference date, and DS1CPOIT indicators in the format 1 DSCB passed to the exit
to arrive at this decision. Command migration does not call this exit, but that
should not be a problem as the user is taking an overt action to migrate the data
set himself. This assumes that you know that the data set was accessed during
a checkpoint.

A single-bit DS1CPOIT was used to indicate both IMSC and physical sequential
MVSC data sets, and the primary intent was to stop migration or any kind of
DFSMSdss processing such as DEFRAG. Remember that we are not talking
about checkpoint data sets but checkpointed data sets. For MVSC, this means
physical sequential data sets that were open during a checkpoint SVC. For
IMSC, this means physical sequential (GSAM) data sets that, as indicated at
open time, would be used during an IMS checkpoint. MVSC and IMSC data sets
have different properties. GSAM data sets can be DEFRAGed, migrated and
recalled, dumped and restored, and copied as long as the amount of data per

22 DFSMS/MVS V1R4 Technical Guide

volume and the device geometry (track size) remain the same before and after
the operations and the data set is not reblocked.

The handling of checkpointed data sets in this way results in several problems:
• DEFRAG was changed to recognize DS1CPOIT and not move the data set to
support the more restrictive MVSC. This resulted in problems for those
installations that were not using MVSC but used IMSC extensively and
depended on DEFRAGing their tape mount management (TMM) DASD
volumes to prevent allocation failures.
• Other DFSMSdss operations such as logical dump, restore, and copy ignore
DS1CPOIT altogether and move the data sets, which can result in a failure
on restarts.
• DFSMSdss physical operations also ignore DS1CPOIT, but they do not
change the amount of data per volume, device geometry (track size), and do
not reblock the data set.
• There was no way to satisfy all customers because of the use of a single bit
for these different types of data sets.

The DFSMS/MVS enhancement enables differentiation between IMSC and MVSC

data sets. It also addresses DEFRAG and all DFSMSdss logical data set
operations.

MVSC will continue to set the DS1CPOIT (checkpointed) indicator and now sets
the DS1DSGU (unmovable) indicator in the format 1 DSCB of the first and current
volume being processed for PS data sets. Open will continue to set DS1CPOIT
for ISMC data sets on the first volume of a multivolume data set. The bit may
not be (and need not be) propagated on secondary volumes for GSAM data sets.

DS1DSGU is new for SMS-managed data sets. It is called the SMS unmovable
indicator. It is treated differently from the DS1DSGU indicator on
non-SMS-managed volumes. In SMS, DS1DSGU indicates that the data set is
unmovable until it is no longer required to be used on a restart. The criterion of
when a data set can be moved is based on when it was accessed during a
checkpoint. Because this time of access is not retained in the format 1 DSCB, it
has to be based on the number of days since last access (last reference date).
New unmovable data sets (PSUs) cannot be created on SMS volumes. The
checks in SMS will continue to be enforced to prevent creation of unmovable
data sets.

The enhanced protection of checkpointed data sets is affected by three DFSMS

components: DFSMSdfp, DFSMSdss, and DFSMShsm.

2.4.1.1 DFSMSdfp Enhancements

Currently checkpoint sets DS1CPOIT in the format 1 DSCB of the first volume of
an SMS-managed physical sequential data set that is open. Checkpoint now
additionally sets DS1DSGU in the format 1 DSCB for these data sets (MVSC data
sets). Checkpoint also updates the date last referenced (DS1REFD) if the current
volume is also the first volume.

O/C/EOV: If open finds DS1DSGU on in a format 1 DSCB for an SMS-managed

data set, it sets DS1DSGU off and issues abend 213-58. Open only issues the
abend if unmovable is specified in either the DCB or JFCB (JCL). Open does not
abnormally terminate if DS1DSGU is on in the DSCB for an SMS-managed data
set.

Chapter 2. DFSMSdfp 23
 Finding DS1DSGU on the format 1 DSCB of a physical sequential SMS data set is
not treated as an error.

2.4.1.2 DFSMSdss Enhancements

When DFSMSdss copies or logically restores a data set, even to a device with
the same track geometry, the extent size and volume layout of the original data
set may be lost. The loss may occur because DFSMSdss attempts to allocate the
copied or restored data set on as few volumes as possible. The IMS
Checkpoint/Restart facilities depend on relative displacement from the beginning
of the first extent on a volume to perform restart on a checkpointed data set.
The MVS Checkpoint/Restart facilities depend on, among other restrictions (like
same volser and extent ___location), the absolute track ___location of restart point to
perform a restart on a checkpointed data set. Neither restart facilities can
tolerate a change in the device type or reblocking of the data.

CONVERTV: DFSMSdss allows a data set that has DS1CPOIT set, with or
without DS1DSGU set, to be converted to NONSMS only if the FORCECP( days )
keyword is specified. If a checkpointed data set is encountered during
conversion to non-SMS and either FORCECP( days ) is not specified or
FORCECP(days ) is specified and the minimum days have not elapsed, message
ADR878E RC102 is issued. If conversion is allowed, DS1CPOIT and DS1DSGU as
well as DS1SMSDS are reset for this data set.

The CONVERTV FORCECP syntax is as follows:

┌── SMS ──┐

CONVERT ──┼─────────┴─────────────────────────────────────┬─
├─ NONSMS ───┬────────────┬─┬─────────────────┬─┤
│ └─ ALLMULTI ─┘ └─ FORCECP(days) ─┘ │
└─ PREPARE ─────────────────────────────────────┘

The FORCECP( days ) parameter specifies that the checkpointed data set resident
on the SMS-managed volume can be converted to non-SMS-managed volumes.
All checkpoint indications are removed from the data set during conversion.
• days is a 1- to 3-digit number in the range 0 to 255. It indicates the number
of days that must have elapsed since the last date referenced before
conversion can take place.

FULL and TRACKS COPY/DUMP: Both full-volume copy/dump and tracks

copy/dump are allowed for a data set that has DS1CPOIT set, with or without
DS1DSGU set. Neither DS1CPOIT nor DS1DSGU is checked or reset.

Full-volume copy/dump of an SMS volume always copies to a device with like

track geometry and always uses the volser of the source volume as the volser of
the target volume.

Tracks copy/dump ought to be reserved as a tool to be used for data repair.

Misuse of tracks copy/dump can cause significant damage to data sets. Not
checking the checkpoint status of a data set has no effect on the potential for
data destruction.

24 DFSMS/MVS V1R4 Technical Guide

COPY DATASET: DFSMSdss allows a data set that has DS1CPOIT set, without
or with DS1DSGU set, to be copied only if the FORCECP( days ) parameter is
specified. If a checkpointed data set is encountered during a data set copy and
either FORCECP( days ) is not specified or FORCECP( days ) is specified and the
m i n i m u m days have not elapsed, message ADR298E is issued. If copy is
allowed, DS1CPOIT and DS1DSGU are reset for the target data set. The settings
of DS1CPOIT and DS1DSGU are not changed for the source data set. Message
ADR297I is issued.

When the target of a data set copy is preallocated and indicated as

checkpointed, the same eligibility criteria are applied to the target data set. If the
target data set is eligible to be replaced, DS1CPOIT and DS1DSGU are reset for
the target, and message ADR297I is issued.

COPY DATASET: The IMS GSAM restart facility can tolerate the relocation of an
extent on a volume. MVS Checkpoint/Restart cannot tolerate the relocation of
any extent needed for restart.

APAR OW08803 changed the DEFRAG function not to move extents of any data
set for which DS1CPOIT was set. DFSMSdss changes the code provided with
APAR OW08803 so that FORCECP( days ) is required to move the extents of a
checkpointed data set if both DS1CPOIT and DS1DSGU are set in the format 1
DSCB. When FORCECP( days ) is not specified or when FORCECP( days ) is
specified and the minimum days have not elapsed, existing message ADR211I is
issued.

DUMP DATASET LOGICAL: A data set that is dumped logically by DFSMSdss

can be restored (in most cases) to a device with different track geometry. Even
when restore is to a device with the same track geometry, DFSMSdss attempts
to allocate the data set in a minimum number of volumes.

The DFSMSdss logical data set dump function is invoked by DFSMShsm for
MIGRATION, BACKUP, and ABACKUP.

An SMS-managed physical sequential data set that has DS1CPOIT set, with or
without DS1DSGU set, is not logically dumped by DFSMSdss unless
FORCECP(days ) is specified. If FORCECP(days ) is not specified or when
FORCECP(days ) is specified and the minimum days have not elapsed, message
ADR298E is issued.

When a checkpointed data set is logically dumped, checkpoint bits DS1CPOIT

and DS1DSGU are not reset, and message ADR297I is issued.

DUMP DATASET PHYSICAL: To be consistent with logical data set dump, users
should check the checkpointed data set indicator before they perform a physical
data set dump. The problem is that a physical data set restore may have as its
source either a physical data set dump or a full volume dump. Therefore, to
ensure that a checkpointed data set is not inadvertently destroyed, there must
be a check on the physical restore side. Dump itself is nondestructive, unless
DELETE is also specified. Since there is no other protection extended to
checkpointed data sets to prevent deletion, it was decided that users should be
able to physically dump with delete, if they so choose.

FULL/TRACKS RESTORE: Both full volume restore and tracks restore are
allowed for a data set that has DS1CPOIT set, with or without DS1DSGU set.
Neither DS1CPOIT nor DS1DSGU is checked or reset.

Chapter 2. DFSMSdfp 25
 A full volume dump of an SMS-managed volume always is restored to a device
with like track geometry and always uses the volser of the source volume as the
volser of the target volume.

Tracks dump/restore ought to be reserved as a tool to be used for data repair.

Misuse of tracks dump/restore can cause significant damage to data sets.

LOGICAL/PHYSICAL DATASET RESTORE: For logical data set restore, when

DS1CPOIT is on, regardless of the setting of DS1DSGU, FORCECP( days ) is
required to restore the data set.

For physical data set restore, FORCECP( days ) is required to restore a

checkpointed data set only if both DS1CPOIT and DS1DSGU are set. IMSC data
sets, those with only DS1CPOIT set, are usable for restart after a physical data
set restore; therefore, FORCECP( days ) is not required for the restore, and
DS1CPOIT is not reset for the restored data set.

DFSMShsm invokes the DFSMSdss function of logical data set restore for
RECALL, ARECOVER, and RECOVER (from BACKUP) and the DFSMSdss function
of physical restore for RECOVER (from DUMP) if DFSMSdss is defined as the
data mover in DFSMShsm.

The DFSMSdss logical data set restore function restores checkpointed data sets
only if FORCECP( days ) is specified. If FORCECP(days ) is not specified or when
FORCECP(days ) is specified and the minimum days have not elapsed, the data
set is not restored and message ADR298E is issued.

When the target of a logical data set restore is preallocated and indicated as
checkpointed, the same eligibility criteria apply to the target. If the target data
set is eligible to be replaced, DS1CPOIT and DS1DSGU are reset for the target.

The DFSMSdss physical data set restore function restores checkpointed data
sets that have DS1DSGU set only if FORCECP( days ) is specified. If
FORCECP(days ) is not specified or when FORCECP( days ) is specified and the
m i n i m u m days have not elapsed, the data set is not restored, and message
ADR298E is issued.

When the target of a physical data set restore is preallocated and has DS1DSGU
set, the same eligibility criteria apply to the target. If the target data set is
eligible to be replaced, DS1CPOIT and DS1DSGU are reset for the target.

For physical data set restore of checkpointed data sets that do not have
DS1DSGU set, FORCECP is not required, and DS1CPOIT is not reset. For
preallocated checkpointed data sets having only DS1CPOIT set, the setting of
DS1CPOIT is taken from the dumped data set rather than the preallocated data
set when the settings are different.

For IMSC data sets that are physically restored, FORCECP(days) is not required,
and the checkpoint indications are not removed from the data sets, whether or
not FORCECP(days) is specified.

RELEASE: The DFSMSdss RELEASE function releases allocated but unused

space from sequential data sets and PDSs.

The RELEASE command operates against SMS-managed physical sequential

checkpointed data sets only when FORCECP( days ) is specified. If

26 DFSMS/MVS V1R4 Technical Guide

FORCECP(days ) is not specified or when FORCECP( days ) is specified and the
m i n i m u m days have not elapsed, message ADR298E is issued.

Checkpoint bits DS1CPOIT and DS1DSGU are reset when allocated but unused
space in a checkpointed data set is released, and message ADR297I is issued. If
a checkpointed data set is eligible for RELEASE but has no allocated but unused
tracks, the checkpoint bits are not reset.

DFSMSdss Enhancement Exit Control: The setting of the FORCECP parameter

and the number of days that must elapse since the date last referenced before a
checkpointed data set can be logically dumped or copied can be overridden by
the DFSMSdss installationwide exit ADRUIXIT, the options exit routine, which is
described in DFSMS/MVS V1R3 Installation Exits (SC26-4908-02).

Installations that support MVSC data sets and/or IMSC data sets and use either
DFSMSdss or DFSMShsm functions must be aware of the support provided in
DFSMS/MVS V1R4 to protect those data sets from inadvertent movement.

Selection of the optimal value for days from function to function and may vary
from application to application, depending on job frequency and use of the
affected data sets.

When a checkpointed data set is not processed because it is ineligible,

DFSMSdss issues an error message to describe the reason the data set is not
processed by the function. DFSMSdss then resumes processing with the next
data set. Any changes in the setting of DS1CPOIT and DS1DSGU occur only for
data sets that have successfully completed all eligibility checks.

2.4.1.3 DFSMShsm Support

One of the main purposes of implementing the DFSMShsm enhancements is to
prevent premature migration of physical sequential SMS-managed checkpointed
data sets.

There is no intent to prevent taking backup copies of a checkpointed data set,

either for normal recovery purposes (RECOVER) or for disaster recovery
purposes (ARECOVER). It is presumed that DFSMShsm users do not initiate a
recover of a data set unless the current copy of the data set, if it exists at all,
has been damaged in some way and is no longer usable. RECOVER is always
command driven, never automatic.

DFSMShsm invokes the services of DFSMSdss for DFSMShsm functions

MIGRATE, RECALL, BACKUP, RECOVER, ABACKUP, and ARECOVER.
DFSMShsm uses the DFSMSdss application interface and passes the request in
the form of a DFSMSdss logical data set dump command (for MIGRATE,
BACKUP, or ABACKUP), a DFSMSdss logical data set restore command (for
RECALL, RECOVER, or ARECOVER), a DFSMSdss physical volume dump
command (for DUMP), or a DFSMSdss physical data set restore command (for
RECOVER).

Checkpoint/Restart Patches: The following patch byte enables users to modify

the number of days that must have elapsed since the date last referenced for a
checkpointed data set (DS1CPOIT set, with or without DS1DSGU set) to be
eligible for migration:

PATCH .MGCB.+112 X′nn′VERIFY(.MGCB.+112 X′05′)

Chapter 2. DFSMSdfp 27
 Note: X′nn′ is the hexadecimal representation for the number of days. The
default is 5.

Checkpoint/Restart Migration: For both command migration and automatic

migration, DFSMShsm determines whether the data set to be migrated is a
checkpointed data set (DS1CPOIT set), with or without DS1DSGU set. A
checkpointed data set is eligible for migration when the date last referenced plus
the number days the data set is to be treated as unmovable are less than or
equal to the current date.
• If the data set is eligible, DFSMShsm builds logical data set dump command
syntax for DFSMSdss including the FORCECP(0) parameter.
• If the data set is not eligible, and the patch for getting ARC0734I messages
for data sets not selected during migration is set, DFSMShsm issues
ARC1245I REASON=90 when space management is running.
• If the data set is not eligible, DFSMShsm issues message ARC1245I
REASON=8 for (H)MIGRATE command migration.

DFSMShsm does not provide an interface whereby a user of the MIGRATE

command can pass the FORCECP parameter in the command syntax. Instead,
the installation must modify the supported patch byte to force the migration.

Checkpoint/Restart Recall: The migration of a checkpointed data set is delayed

by at least the unmovable days in the migration path. Therefore any
checkpointed data set that was migrated can be recalled without any additional
delay. For RECALL, DFSMShsm calls DFSMSdss with the parameter
FORCECP(0) in the command syntax for DFSMSdss logical data set restore.

DFSMSdss needs FORCECP(0) to indicate that the restore should be allowed to

proceed.

Checkpoint/Restart BACKUP and ABACKUP: The backup of a data set with

BACKUP or ABACKUP is not delayed for a checkpointed data set. DFSMShsm
invokes DFSMSdss with FORCECP(0) in the command syntax to allow backup of
any checkpointed data sets.

Checkpoint/Restart Volume Dump: Because DFSMSdss does not distinguish

checkpointed data sets during full volume dump, a keyword is not required on
the DFSMSdss invocation of a dump to cause a checkpointed data set to be
dumped.

Checkpoint/Restart RECOVER and ARECOVER: The recovery of a data set, with

RECOVER or ARECOVER, is not delayed for a checkpointed data set.
DFSMShsm invokes DFSMSdss with FORCECP(0) in the command syntax to
allow recovery, including replacement, of any checkpointed data sets.

2.4.1.4 Enhanced Checkpoint/Restart Migration and Coexistence

DFSMSdfp O/C/EOV and DFSMSdss are affected by the migration process if
other DFSMS/MVS releases exist in the installation.

O/C/EOV Migration and Coexistence: The setting of DS1DSGU in the format 1

DSCB (product sensitive programming interface) of an SMS-managed physical
sequential data set is incompatible with previous releases of DFSMS/MVS and
DFP. Therefore, O/C/EOV has toleration PTFs to previous DFSMS/MVS releases
and DFP 3.3.0.

28 DFSMS/MVS V1R4 Technical Guide

 DFSMSdss Migration and Coexistence: DFSMSdss does not supply compatibility
or toleration PTFs to any previous releases. Previous releases do not honor the
FORCEP(days) parameter.

A DEFRAG APAR, which allows extents to be moved when DS1CPOIT is set, will
be maintained for previous levels of DFSMSdss.
Note: OEM vendor products or user applications using the format-1 DSCB fields
should be aware of the use of the DS1DSGU field introduced by the enhanced
protection of checkpointed data sets.

2.5 Optical Access Method SMF Recording Enhancement

Customers must be able to perform the following activities:
• Performance monitoring, analysis, and tuning
• Capacity planning, including:
− CPU utilization and growth planning
− Storage subsystem utilization and capacity planning for tape, DASD, and
optical storage subsystems
− Removable media capacity planning related to removable media such as
tape and optical
− Accounting for information systems use and chargeback to end-user
departments for the use of information systems resources.

The SMF recording enhancement assists customers who are using OAM in their
performance monitoring, analysis, tuning, and capacity planning activities.

This enhancement introduces a new SMF record for OAM at the OAM
programming interface level (the OSREQ macro interface). In an ImagePlus
environment, these records are used to account for the OAM portion of
document image processing.

The MVS system operator or system programmer can dynamically select the
OAM SMF record subtypes to be recorded.

The OAM SMF record enables customers to collect substantial statistical

information about OAM usage.
Note: The OAM SMF recording enhancement applies only for the 3995 optical
library dataserver. IBM ATL devices such as the 3494 and 3495 are not affected
by this OAM SMF recording enhancement.

OAM records SMF records in the SMF data set to account for OAM activity.
Each OAM SMF record contains three sections:
• Standard 48-byte SMF record header
• 112-byte OAM product section
• Variable length OAM data section

Table 2 lists the OAM SMF record subtypes that OAM SMF record type 85 (x ′55′)
supports.

Chapter 2. DFSMSdfp 29
 Table 1. OAM SMF Record Subtypes
Size Description
1 324 OSREQ access
2 324 OSREQ store
3 324 OSREQ retrieve
4 324 OSREQ query
5 324 OSREQ change
6 324 OSREQ delete
7 324 OSREQ unaccess
32 336 OSMC storage group processing
33 336 OSMC DASD space management
34 336 OSMC optical disk recovery utility
35 336 OSMC MOVEVOL utility
36 296 OSMC single object recovery utility
37 184 OSMC library space management
64 256 LCS optical drive vary online
65 256 LCS optical drive vary offline
66 256 LCS optical library vary online
67 256 LCS optical library vary offline
68 284 LCS optical cartridge entry
69 284 LCS optical cartridge eject
70 284 LCS optical cartridge label
71 284 LCS optical volume audit
72 284 LCS optical volume mount
73 284 LCS optical volume demount
74 variable (min. 380, max. 32,744) LCS optical write request
75 380 LCS optical read request
76 380 LCS logical delete request
77 variable (min. 380, max. 32,744) LCS optical physical delete request
78 variable (min. 380, max. 32,744) LCS object tape write request
79 380 LCS object tape read request
87 228 LCS Object tape volume demount (OAM usage)

2.5.1 Start Time and End Time Accuracy

Each OAM SMF record has a function start time and end time in the common
OAM product section. The start time and end time are in System/390 store clock
(STCK) instruction format.

Every attempt is made to obtain the start and end time of the OAM function as
soon and as close as possible so that the elapsed time of the function includes
as much OAM processing time as possible.

30 DFSMS/MVS V1R4 Technical Guide

2.5.1.1 OSREQ Macro TTOKEN Keyword
A new optional keyword, TTOKEN, is added to the OSREQ macro. The OSREQ
macro is the general-use programming interface provided with OAM. The new
keyword is valid on all forms of the OSREQ macro:
• List (MF=L)
• Execute (MF=E)
• Modify (MF=M)

The format of the new keyword is:

{TTOKEN = ¬ tracking-token | (tracking-token-pointer)]}

where tracking-token specifies a 16-byte area containing a tracking token. The

contents of the 16-byte tracking token may be anything. The tracking token
supplied on the OSREQ macro with the TTOKEN keyword is placed in OAM SMF
record subtypes 1 through 7. If no tracking token is supplied on the OSREQ
macro, the tracking token field in record subtypes 1 through 7 will contain binary
zeros.

A new reason code is returned in register 0, following the OSREQ macro, if the
tracking token is contained in a virtual storage area to which the application
program does not have both fetch and store authorization.

2.5.1.2 Invocation
Before activating OAM SMF recording following an MVS IPL, the OAM subsystem
identification (OAM) must be defined in the active SMF PARMLIB (SMFPRMxx).
If the OAM subsystem identification has not been defined to SMF, add the
following statements to the SMFPRMxx:
SUBSYS(OAM,TYPE(85))
and then activate it by entering the following MVS operator SET command:
SET SMF=xx

If the OAM subsystem identification has been defined to SMF, the MVS system
operator or MVS systems programmer can dynamically change OAM SMF
recording, using one of the following methods:
• Update the SMF PARMLIB member to include the OAM SMF record
subtypes, for example:
SUBSYS(OAM,TYPE(85(2:3)))
and activate the SMFPRMxx by entering the following MVS operator SET
command:
SET SMF(xx)
• Update the SMF options dynamically by entering the following MVS operator
SETSMF command:
SETSMF SUBSYS(OAM,TYPE(85(4:6)))

For additional information about the MVS operator SET SMF and SETSMF
commands, see the MVS/ESA System Commands Reference , (GX22-0015-03).

Chapter 2. DFSMSdfp 31
2.6 Program Management 3 (PM3)
DFSMS/MVS provides program management services to create, load, modify,
list, read, transport, and copy executable programs. DFSMS/MVS 1.1 introduced
the program management binder and the program management loader. The
binder extends the services formerly provided by the MVS/DFP linkage editor
and batch loader. These enhancements include support for an executable unit
called a program object , which includes all of the functions of a load module,
with additional functional and usability improvements. The loader adds to the
capabilities of program fetch and can load both program objects and load
modules into storage for execution.

2.6.1 Program Management - Background

The linkage editor, batch loader, and program fetch programs are now stabilized
and will not be enhanced. In DFSMS/MVS 1.1 they were replaced by the binder
and loader, which compatibly include all functions of their predecessors and
incorporate many enhancements and new functions. A more powerful binding
algorithm has been added to the binder, greatly increasing the flexibility of
DFSMS/MVS program management in support of the language products.

In addition to support for the load module, the binder introduced a new
executable unit, the program object. The program object removes all structural
limitations that have long restricted the load module. Flexibility in the design
and structure supports ongoing enhancements.

Program objects have the following base features:

• Program object design allowed for the removal or increase of most size
restrictions, including maximum text size (now 1 GB) and number of control
sections (now unlimited).
• Because program objects reside exclusively in PDSEs, they can take
advantage of that library technology and its many advantages.
• The program object structure is generalized and extendable. It has changed
and will continue to change with each PM release.
• Program objects, physically organized as 4K fixed blocks, can be page
mapped into storage (using DIV).
• Program objects support long names (up to 1K).
• Program objects contain many of the same enhancements supported in the
new object file, generalized object file format (GOFF), which is currently
created by the High Level Assembler (as well as the Binder as an
intermediate structure). This includes support for C and C++ writable
static.
• Program objects contain multiple classes of text, distinguished by attributes
that control binding and loading characteristics and behavior. Classes are
central to C and DLL support.
− There are two types of classes - text (byte stream) and nontext
(recordlike, IDR, ADATA)
− The separate attributes assigned to each class include:
- LOAD - The class is brought into memory at the time the module is
loaded (typical case today).

32 DFSMS/MVS V1R4 Technical Guide

 - DEFERRED LOAD - The class is prepared for loading, but not
instantiated until requested (new in PM3).
- NOLOAD - The class is not loaded with the program, that is, it is
nontext.
- RMODE 24/ANY - Indicates placement of segments within virtual
storage
− A section is the smallest unit that users can manipulate (replace, delete,
order). Each section may supply contributions to one or more classes.
− Classes are bound into independently loadable segments . A segment
contains classes with compatible attributes. A program object may have
multiple segments.
− The loading characteristics of the class (and segment) determine the
placement of the segment in virtual storage. Multisegment program
objects can be loaded into noncontiguous areas of virtual storage, for
example, when bound with the RMODE(SPLIT) option.
− Program objects contain a class of data specifically intended for users to
save associated or application data (ADATA). It is not loadable
(NOLOAD). This data includes source statements, debugging tables,
user information, history data, and documentation. It is accessible
through the binder API.

2.6.2 PM3 Enhancements for DFSMS/MVS V1R4

The basic functions of the binder were not modified in PM3; that is, the binder
binds sections to create modules and saves them in libraries or prepares them
for immediate execution. In addition, the binder API continues to provide
support for calling programs to access data and to construct, bind, or copy
modules.

PM3 addresses some current problems in support of C and C++. It simplifies

the creation of an executable module and extends support for OpenEdition.
Specifically PM3 includes:
• A conversion function added to the binder to process XOBJ structures built
by the C and C++ compilers. The output from the binder on this path is a
program object that must be stored in a PDSE. This removes the need for
the LE/370 Prelinker in those cases where a PDSE is specified as the target
load library (that is, SYSLMOD).
• Enhanced binder and loader to support OS/390 dynamic linking and DLLs.
This enhancement provides for user-controlled special handling of external
references in modules, to allow their resolution when loading the referenced
module explicitly or at the time the module is referenced, that is, it supports
forms of dynamic loading.
• Support for the above two enhancements has also been included in
OpenEdition and the C89 shell command.
• The loader is enhanced to support DLLs and deferred loading, as well as the
OS/390 R4 Dynamic LPA functions, extending this support to PDSEs and all
program objects, including DLLs.

Chapter 2. DFSMSdfp 33
34 DFSMS/MVS V1R4 Technical Guide
Chapter 3. DFSMSdfp VSAM

In this chapter we describe the following enhancements to VSAM delivered in

DFSMS/MVS V1R4:
• VSAM system-managed buffering
• VSAM fast load
• Update of the VSAM last reference date at close
• VSAM RLS KSDS extended addressability
• Data class support for more VSAM attributes
• DCOLLECT
Note: Now is probably a good time to review all performance-oriented
parameters that can be specified in the storage class, such as BIAS, MSR, and
SDR. You should also review the characteristics of extended format (EF)
sequential and VSAM KSDS data sets and their definition, using ISMF data class
panels.

Because of the additional usability characteristics of data sets defined in EF,

more and more function is being delivered through EF data sets. Data set
allocation and performance are directly affected by approprialely setting the
storage class values.

3.1 VSAM System-Managed Buffering

VSAM system-managed buffering (SMB) enables VSAM to determine the
optimum number of index and data buffers, as well as the type of buffer
management (sequential or direct). Where appropriate a switch to local shared
resource (LSR) buffering occurs automatically without changes to JCL, as are
required now to use Batch LSR processing.

Performance improvements could be dramatic, particularly where defaults for

the number of buffers are used, or a switch from nonshared resource (NSR)
processing to LSR is made.

SMB is available under the following conditions:

• The VSAM data set must be in EF
EF data sets are required to be system managed.
• ACB MACRF must be NSR
The ACB MACRF parameter must not contain LSR, GSR, RLS, ICI, AIX, or
UBF.
• Record_Access_Bias must be set to SYSTEM in the data class, or the JCL
AMP parameter ACCBIAS must be set to SYSTEM, SO, SW, DO, or DW. For a
description of these parameter values, see 3.1.1, “Record Access Bias
Specification” on page 36.

VSAM is guided on how to optimize buffers based on the Record_Access_Bias

parameter specified in the data class or on the JCL AMP parameter. The
manner in which buffers are processed refers to buffer management continuing

 Copyright IBM Corp. 1997 35

 as NSR, or being changed internally to LSR to take advantage of LSR
capabilities, which may include the use of hiperspace.

The order of precedence for specifying values is: JCL over data class, with data
class having precedence over specifications in the ACB. The MACRF values of
SEQ, DIR, and SKP are used with a specification of SYSTEM to determine how
buffers will be acquired. Also used with SYSTEM are the storage class values
for BIAS and/or MSR, should they be specified.

3.1.1 Record Access Bias Specification

Record Access Bias is specified on either the JCL DD statement or in the data
class:
• JCL
The AMP parameter can be used to specify access bias with a subparameter
of ACCBIAS. This subparameter can have one of six specifications:
USER Bypass SMB
SYSTEM Force SMB and let the system determine the buffering technique
according to the ACB MACRF and storage class specifications
SO SMB with sequential optimization
SW SMB weighted for sequential processing
DO SMB with direct optimization (forces switch to LSR)
DW SMB weighted for direct processing
• Data class
A new subparameter for a DATA SET NAME TYPE of EXT is REC_ACC_BIAS.
Either USER or SYSTEM can be specified for REC_ACC_BIAS.

Specifying the type of Record Access Bias through the AMP parameter in JCL
will override anything specified in the data class for this parameter. If nothing
has been specified for this parameter, the default is USER. USER indicates that
VSAM will continue to use buffers as it now does without SMB. Record Access
Bias is ignored if the data set is not in EF.

SMB either weights the buffer handling toward sequential or direct processing or
optimizes the buffer handling for sequential or direct processing. Weighting
means that when the Record Access Bias specifies SW (sequential weighted)
most buffers will be used to support sequential processing but some will be
reserved for index buffers to help any direct processing. Conversely when the
Record Access Bias specifies DW (direct weighted), most buffers will be used to
support fast direct access to the data, with relatively few buffers reserved for any
sequential processing that might occur. The manner in which buffer selection
may be affected by the user program is specified by the use of the ACB MACRF
parameters of SEQ, SKP, and DIR and the use of the storage class BIAS and/or
MSR values (see Table 2 on page 37).

A value of SYSTEM (including SO, SW, DO, or DW for ACCBIAS) specifies that
VSAM is to determine the number of buffers to obtain for the data set when NSR
processing is used. When NSR processing is specified or defaulted, and VSAM
chooses DO (direct optimized) as the most appropriate type of access, the
buffering technique changes from NSR to LSR. LSR is also used if direct
optimization is forced by specifying ACCBIAS=DO on the AMP parameter.

36 DFSMS/MVS V1R4 Technical Guide

 When SMB determines that LSR buffer management is to be used, VSAM also
determines, by default, the number of virtual storage buffers to use.

Table 2. ACB MACRF Parameters and Storage Class BIAS

ACB MACRF Parameters Storage Class BIAS
SEQ DIR Both None
MACRF=DIR DW DO DO DO
MACRF=SEQ (default) SO SW SO SO
MACRF=SEQ and SKP SO SW SW SW
MACRF=SKP DW DW DW DW
MACRF=(DIR,SEQ) or (DIR,SKP) or SW DW DW DW
MACRF=(DIR,SEQ,SKP)
Note: DO=direct optimized, DW=direct weighted, SO=sequential optimized, SW=sequential weighted

3.1.2 ISMF Data Class Panels

Several ISMF panels have been updated for the data class to support
RECORD_ACCESS_BIAS = S (System) or U (User).

Neither SO, SW, DO, nor DW are specified in the data class as these are
determined from a consideration of the storage class BIAS and/or MSR values
and the ACB MACRF values as shown in Table 2.

3.1.3 Buffers Used for Direct Optimization

If DO is set, SMB converts NSR buffering to LSR buffering, in which case three
new optional AMP parameters can be specified to tell the LSR buffer manager
how to handle the processing of the buffers. The three new subparameters are:
SMBVSP specifies the amount of virtual storage to obtain for buffers when
opening the data set
SMBHWT used to allocate hiperspace buffers based on a multiple of the number
of virtual buffers that have been allocated
SMBDFR allows the user to specify whether writing the data from a buffer to
disk can be deferred until the buffer is required for a different request
or the data set is closed. A CLOSE TYPE=T will not write the buffers
to disk when LSR processing is used for direct optimization.

3.1.3.1 Virtual Buffers for Direct Optimization

The two new AMP subparameters, SMBVSP and SMBDFR, can be used to tell
the LSR buffer management how to handle the processing of the virtual buffers.

The amount of virtual buffer space to be acquired when opening the data set for
LSR processing can be specified with the SMBVSP subparameter of AMP. The
format is:

SMBVSP=nnK

where nn is 1 to 204800
or
SMBVSP=nnM

Chapter 3. DFSMSdfp VSAM 37

 where nn is 1 to 2048

The SMBVSP parameter can be used to override the default buffer space to be
obtained, which generally speaking is calculated assuming 20% of the data will
account for 80% of the accesses. The buffer space acquired is split across two
LSR pools; one for the index and one for the data.
Note: The value specified is the total amount of virtual storage that can be
addressed in a single address space. It does not take into consideration the
storage required by the system or the access method.

When direct optimization is used, the write processing of modified buffers can be
deferred until the data set is closed or the buffer is required for a different
request. The SMBDFR subparameter of AMP is used to specify deferred write.
The format is:

SMBDFR=Y|N

where Y is the default for SHAREOPTIONS (1,3) and (2,3), and N is the default for
SHAREOPTIONS (3,3), (4,3), and (x,4)

3.1.3.2 Hiperspace Buffers for Direct Optimization

Direct optimization converts NSR buffering to LSR buffering, which facilitates the
use of additional buffers that are kept in hiperspace.

The amount of hiperspace to be used for LSR buffers can be specified with the
SMBHWT subparameter of AMP. The value specified for SMBHWT is used as
the hiperspace weighting factor. The format is:

SMBHWT=nn

where nn is a whole decimal number between 1 and 99

The hiperspace buffer size will be a multiple of 4096 (4K). These buffers may be
allocated for the base data component of the sphere.

If the control interval (CI) size of the database component is not a multiple of 4K,
both virtual space and hiperspace are wasted. In addition, excessive use of this
facility will have a negative affect on performance for both the system and the
user program. Overhead is involved in locating a buffer for the user program
that may not be in hiperspace storage because of system requirements.

The SMBHWT value is not a direct multiple of the number of virtual buffers that
are allocated to the resource pool but act as a weighting factor for the number of
hiperspace buffers to be established. If SMBHWT is not specified, hiperspace
will not be used.

3.1.4 Location of Buffers and Control Blocks

Although SMB allows for the specification of buffer handling techniques, it does
not allow the user to specify the ___location of buffers and control blocks. A new
JCL AMP parameter, RMODE31, is provided to allow the user to specify the
___location of buffers and control blocks. It will override any specification for the
ACB macro. This capability is necessary when there are a large number of data
sets open to the job and/or the number of buffers may cause a storage shortage
when kept in 24-bit addressable storage. The values that can be specified for
RMODE31 with the JCL AMP parameter or the ACB are:

38 DFSMS/MVS V1R4 Technical Guide

 CB - Control blocks above 16 MB
NONE - control blocks and buffers below 16 MB
BUFF - Buffers above 16 MB (this is the default)
ALL - Control blocks and buffers above 16 MB

The RMODE31 parameter that is currently specifiable in the ACB will be merged
with, or overridden by, the new JCL RMODE31= AMP subparameter.

The merge of the AMP=(′RMODE31=...′) and the ACB RMODE31= keyword is

summarized in Table 3. The DEFAULT value results when RMODE31 is not
specified in JCL, Data Class RECORD_ACCESS_BIAS is not USER, and JCL AMP
ACCBIAS is not USER.

Table 3. RMODE31 Selection

A M P = ( ′RMODE31=...′) ON JCL
ACB RMODE31 OPTION Control
NONE BUFF ALL DEFAULT
block
NONE NONE BUFF Control ALL BUFF
block
BUFF NONE BUFF Control ALL BUFF
block
Control block NONE BUFF Control ALL ALL
block
ALL NONE BUFF Control ALL ALL
block

Note: The use of the JCL AMP RMODE31= parameter is valid only on systems
with DFSMS/MVS V1R4 installed. A JCL error will be returned for earlier
releases.

3.1.5 LSR Resource Pools

Currently, when using local shared resources, each address space may have up
to 16 index resource pools and 16 data resource pools independent of other
address spaces. The SHRPOOL parameter is used to indicate which resource
pool you are using. The SHRPOOL value must be a number between 0 and 15.

With DFSMS/MVS V1R4 the number of LSR pools has been increased from 16 to
256. The value specified for SHRPOOL will now be a number between 0 and 255.
This will cause changes in the BLDVRP, DLVRP, ACB, GENCB ACB and MODCB
ACB macro instruction SHRPOOL= values.

3.2 VSAM Load Enhancements

Changes have been made to improve the performance of loading an EF KSDS by
reducing the number of I/O requests required to write the data.

As illustrated in Figure 5 on page 40, the conditions under which this

enhancement take effect are:
• The KSDS must be in EF.
• The data set is defined with the SPEED parameter.
• SMB is requested in the data class or on the AMP parameter.

Chapter 3. DFSMSdfp VSAM 39

 For more information about SMB, see 3.1, “VSAM System-Managed Buffering”
on page 35.

Figure 5. VSAM Loading Enhancements

The number of data buffers is optimized for load mode processing, which is
sequential; system-managed buffers adjust to the load mode requirements by
determining the current state of the data set. Sufficient buffers are acquired
such that each control area (CA) is written with a single I/O request and full
overlap of I/O and processing occurs. Before this release at least two I/O
requests would have been needed for each CA, more if the data set was defined
with the FREESPACE parameter specifying one or more free CIs per CA.

The index component (including the sequence set) should be updated only once
per data CA area during load. Before this release the index was updated
multiple times per CA. This change should improve load performance
considerably.

Performance should also improve because the entire CA is written to cache with
a single I/O request, and a device end is returned before all of the data is written
to DASD. Even in cases where the algorithms bypass cache, performance
should improve as the entire CA is written with a minimum of disk rotations.

3.2.1 Invocation
The following specifications are required for the load enhancements to take
effect:
• System managed buffering
• SPEED specified in the IDCAMS DEFINE command

3.2.2 Interfaces
The interfaces used for this support are the IDCAMS DEFINE command and
either the data class or the JCL AMP parameter.

40 DFSMS/MVS V1R4 Technical Guide

3.3 Update VSAM Last Reference Date at CLOSE
The last reference date (LRD) is now updated in the format-1 DSCB on the first
volume for the base data component of a VSAM sphere when the data set is
closed in addition to open. The LRD is updated when the close is the last close
for the data set. The current date must be greater than the date on which the
data set was opened. It must also be greater than the DS1REFD date in the
format-1 DSCB.

Before to DFSMS/MVS V1R4, the LRD for VSAM data sets was updated at open
time. Thus there was always the potential for VSAM data sets that had been
open for many days to be migrated unnecessarily.

Updating the LRD only at OPEN time is of great concern to some installations
whose transaction-based systems may have a large number of data sets and/or
databases open for many days or even weeks. When the transaction system is
stopped, all of the data sets are eligible for migration by DFSMShsm according
to the LRD set at OPEN time. Depending on the number and size of the data
sets, the time to recall them could be long, thus delaying the time to bring up the
transaction system again.

As illustrated in Figure 6 on page 42, updating the LRD at CLOSE time prevents
the immediate migrate and recall activity and brings VSAM inline with updating
the LRD for non-VSAM data sets at CLOSE time.

For non-RLS VSAM, the IDATMSTP routine is called during the open of the data
set to retrieve a return code that specifies whether or not the date in the VTOC
is to be changed. VSAM keeps this information until the data set is closed. For
VSAM RLS, date stamp processing is always performed. Data stamp processing
for close compares the date on which the data set was opened with the date on
which the data set is closed to determine whether the date has changed.

3.4 VSAM RLS KSDS Extended Addressability

When VSAM RLS support was announced with DFSMS/MVS V1R3, it did not
support VSAM KSDS extended addressability (EA) format. In other words VSAM
RLS data sets retained the 4GB architectural limit for data set size imposed by
using the 4-byte field for the relative byte address (RBA) addressing mechanism.
DFSMS/MVS V1R4 removed this restriction by supporting RLS EA for VSAM
KSDSs, and hence VSAM RLS now supports VSAM KSDSs up to the current
multivolume limit of 59 DASD volumes.
Note: CICS was the first subsystem to exploit VSAM RLS, although batch VSAM
programs could also access these data sets with read integrity. With
DFSMS/MVS V1R4, DFSMShsm will use VSAM RLS for its control data sets.

DFSMS/MVS V1R4 supports EA VSAM KSDS only. It does not support other
types of VSAM data sets for EA.

To use VSAM RLS the installation must have implemented a Parallel Sysplex
environment. A coupling facility is therefore a prerequisite before VSAM RLS can
be implemented. For a brief overview of a Parallel Sysplex and VSAM RLS, see
Appendix A, “Parallel Sysplex and VSAM Record Level Sharing” on page 139.

Chapter 3. DFSMSdfp VSAM 41

Figure 6. VSAM Last Reference Date at CLOSE. VSAM n o w updates the last reference date field at CLOSE time,
thereby reducing the chance that a data set that has been open for many days would be migrated, only to be
immediately recalled.

3.4.1 Objectives and User Requirements

VSAM RLS KSDS extended addressability satisfies these two requirements:
• A VSAM KSDS should be capable of being shared with integrity for
concurrent updates from multiple processors in a parallel sysplex.
• The VSAM KSDS being shared should be capable of being much larger than
4GB.

This also satisfies the requirement of supporting the expansion of a compressed

extended format VSAM KSDS into a decompressed VSAM RLS extended
addressability KSDS that can be larger than 4GB.

3.4.2 Functional Characteristics

To accommodate data set sizes greater than 4GB changes have been made to
VSAM RLS code from RBA fields to CI-number.

As in VSAM non-RLS extended addressability, a VSAM RLS extended

addressability KSDS must have its data class defined with:

Data Set Name Type =EXT

If Ext =P or R
Extended Addressability =Y

Just as with VSAM non-RLS EA, an application can use the TESTCB macro to
test for extended addressability with VSAM RLS EA with, for example:

TESTCB ACB=EFKSDS,ATRB=XADDR

42 DFSMS/MVS V1R4 Technical Guide

 Note: VSAM RLS (extended addressability or non-extended addressability) does
not allow access to data through the RBA. RPL RBA or XRBA for GET and PUT
are ignored without errors. However, control block manipulation macros
GENCB, MODCB, and SHOWCB still work, even if the RBA or XRBA parameter is
coded on the RPL.

Products that rely on the expanded data set size to determine the most efficient
method of processing the data or return the logical data set size to the user can
still do so for both EA and non-EA data sets, using an externally callable service.
For example:

SHOWCB ACB=EAKSDS,
FIELDS=(XAVSPAC,XENDRBA,XHALCRBA)

could be used to return the amount of space, the last RBA value of the data set,
and the high-used RBA of the data set. The field values are returned as 8-byte
values instead of 4-byte values.

3.4.3 Migration and Coexistence

The following statements should help you understand the compatibility issues
between DFSMS/MVS V1R3 and DFSMS/MVS V1R4 systems for processing a
KSDS and between EA and non-EA versions of a KSDS:
1. The format of a non-EA extended format KSDS created on either
DFSMS/MVS V1R3 or DFSMS/MVS V1R4 is identical. These extended format
KSDSs will continue to have RBA values in the catalog and index sequence
set, whether or not they are processed in RLS mode.
2. A VSAM Extended Addressable KSDS has an index that is incompatible with
an extended format KSDS index without EA. As a result, a KSDS allocated
as EA capable and processed in RLS mode from a DFSMS/MVS V1R4 system
cannot be processed with VSAM RLS on a DFSMS/MVS V1R3 system. The
DFSMS/MVS V1R3 code already supports this, failing OPEN with an IEC161I
005-731 message because it does not support VSAM RLS KSDS EA.

Use IDCAMS REPRO to migrate from non-EA KSDSs to EA KSDSs or vice versa.
You can also use IMPORT/EXPORT as long as the data set is not larger than
4GB. It is not possible to use ALTER to change an existing extended format
non-EA format KSDS into an EA format KSDS.

3.4.4 Software Prerequisites

There are two dependencies:
• DFSMS/MVS V1R4
• MVS/ESA SP Version 5.2 or OS/390 R4 and later

Only data sets on SMS-managed volumes are eligible for EA because the
mechanism for obtaining EA is the specification of a Data Set Name Type of EXT
and Extended Addressability set to Y in the data class.

Chapter 3. DFSMSdfp VSAM 43

 3.4.4.1 Program Evaluation
Whether KSDSs with EA are used in RLS mode or not, programs accessing them
must be investigated to determine whether they can be greater than 4GB.
Note: If an extended addressable KSDS is used in RLS mode, RBA processing
is not allowed.

3.4.4.2 JCL Specification

To create a KSDS with EA, you can use either the LIKE or DATACLAS data
definition keywords. If the LIKE keyword is specified, the data set used as the
model must already be a KSDS with EA. If the DATACLAS keyword is specified
or assigned by the ACS routines, the data class must include a Data Set Name
Type of EXT with Extended Addressability set to Y.

3.4.5 Configuration Specifications

To support VSAM RLS EA, the data set must reside on a DASD subsystem that
supports the extended platform and thus extended format data sets.

VSAM RLS KSDS EA will operate on all processors supported by MVS/ESA 5.2 or
later. Coupling Facility hardware in the System/390 Sysplex is required. See
Appendix A, “Parallel Sysplex and VSAM Record Level Sharing” on page 139
for further information.

3.5 Data Class Supports More VSAM Attributes

You can define most data set attributes, such as the organization, record format,
and space required, in the data class by using ISMF. This allows data sets to be
allocated via JCL or TSO ALLOCATE or dynamic allocation. The main
exceptions have been the attributes of BWO, LOG, and LOGSTREAMID made
available with DFSMS/MVS V1R3, and the SPANNED/NONSPANNED attribute
available since 1974. If users needed these attributes, they had no option but to
specify them using the IDCAMS DEFINE command or the IDCAMS ALTER
command.

DFSMS/MVS V1R4 has enhanced ISMF and the Data Class Application to allow
these attributes to be defined in the data class.

3.5.1 User Requirements

Users have requested that all VSAM data set definition attributes be available
from a data class. DFSMS/MVS V1R4 meets this requirement now with the
inclusion of BWO, LOG, LOGSTREAM ID, and the SPANNED/NONSPANNED
attributes in the data class. Hence it is possible to define any VSAM data set
with its required attributes through JCL, as illustrated in Figure 7 on page 45.
This should be considered to be the preferred way of defining VSAM data sets in
the future. In many cases a separate IDCAMS DEFINE or IDCAMS ALTER step
can be eliminated.

IF JCL and the data class are used to set up a VSAM data set, attributes
conflicting with each other, or with the data set definition, can be caught when
the data set is created rather than when the data set is first opened.

44 DFSMS/MVS V1R4 Technical Guide

Figure 7. VSAM Data Class. The Data Class n o w allows m o r e attributes to be specified, including the SPANNED
attribute.

3.5.1.1 ISMF Support

ISMF has been modified to accept the BWO, LOG, LOGSTREAM ID, and
SPANNED/NONSPANNED attributes in the Data Class Application:
• BWO(TYPECICS|TYPEIMS|NO|blank)
BWO is an optional attribute in the data class. It should be specified if
backup-while-open (BWO) processing is allowed for the VSAM sphere. BWO
applies only to SMS VSAM data sets and cannot be used with TYPE(LINEAR).
− TYPECICS (abbrev. TYPEC)
Use TYPECICS to specify that BWO is to be performed for CICS VSAM
file control data sets, to allow forward recovery using a BWO copy. The
LOG attribute determines whether CICS will use this attribute from the
data class or use values from the file control table (FCT).
− TYPEIMS (abbrev. TYPEI)
Use TYPEIMS to specify that BWO is to be performed for IMS VSAM data
sets.
− NO
Use NO when BWO does not apply to the VSAM cluster.
− blank (default)
For ISMF, blank is the same as NO (that is, BWO is not used). When
blank is defaulted, the attribute is not saved in the catalog.
• LOG(NONE|UNDO|ALL|blank)
LOG is an optional attribute in the data class. It establishes whether the
sphere to be accessed with VSAM RLS is recoverable or nonrecoverable. It

Chapter 3. DFSMSdfp VSAM 45

 also indicates whether a forward recovery log is available for the sphere.
LOG applies to all components of the sphere.
The LOG attribute determines whether CICS uses the FCT or the values in
the catalog. If the LOG attribute exists in the catalog, CICS uses the LOG,
LOGSTREAM ID, and BWO attributes from the catalog. If the LOG attribute
does not exist in the catalog, CICS uses the FCT definitions for LOG,
LOGSTREAM ID, and BWO.
− NONE
Indicates that neither an external backout nor a forward recovery
capability is available for the sphere accessed in RLS mode. If NONE is
specified, RLS considers the sphere to be nonrecoverable.
− UNDO
Specifies that changes to the sphere accessed in RLS mode can be
backed out by using an external log. RLS considers the sphere to be
recoverable when LOG(UNDO) is specified, but forward recovery is not
possible.
− ALL
Specifies that changes to the sphere accessed in RLS mode can be both
backed out and forward recovered by using external logs. RLS considers
the sphere recoverable when you use LOG(ALL).
− blank (default)
When blank is defaulted, CICS use the FCT definition for non-RLS
processing. OPEN is not allowed for RLS access.
• LOGSTREAM ID(dsname|blank)
LOGSTREAM ID is an optional attribute in the data class. It is used to name
the CICS forward recovery log stream. It applies to all components of the
VSAM sphere.
− dsname
The name of the forward recovery log stream. The data set name (the
logstreamid) has a maximum length of 26 characters, including
separators.
− blank (default)
When this attribute is not specified in the data class, SMS fails dynamic
or JCL allocation of data sets with the LOG(ALL) attribute. On an AMS
DEFINE, after merging data class attributes, AMS fails the DEFINE if the
LOG(ALL) attribute is specified or derived and LOGSTREAMID is not
specified or derived.
• SPANNED|NONSPANNED|blank
SPANNED|NONSPANNED is an optional attribute in the data class. It
specifies whether a data record is allowed to cross CI boundaries. These
parameters have been a part of standard VSAM record management since
VSAM was first delivered in 1974 and are valid for SMS-managed and
non-SMS-managed data sets.
− SPANNED

46 DFSMS/MVS V1R4 Technical Guide

 Specifies that if the length of a data record is larger than a CI, the record
will be in more than one CI. Thus VSAM can select a CI size that is
optimal for the DASD.
− NONSPANNED
Indicates that the record must be in one CI. VSAM selects a CI size that
accommodates the maximum record size.
− blank
Defaults to NONSPANNED. Hence records are not allowed to expand
beyond the boundary of a single CI.

3.5.1.2 Attribute Precedence

AMS merges the BWO, LOG, LOGSTREAM ID, and SPANNED/NONSPANNED
attributes from the data class if not specified on DEFINE.

When a data set is created, SMS applies the data class values for the attributes
only if they apply to the data set. For example, none of these attributes is
compatible with TYPE(LINEAR) and therefore none is applied to linear data sets.

3.5.1.3 Impact on the catalog

The BWO, LOG, and LOGSTREAM ID attributes are saved in the RLS cell of the
data set in the catalog. The RLS cell is not created if these attributes are not
specified or are not applicable. The cell is also not created for non-VSAM data
sets and non-SMS data sets.

3.5.2 Migration and Coexistence

Upward compatibility is maintained.

The LOGSTREAM ID value can be up to 26 characters and increases the length

of the data class construct. Toleration PTFs are required in ISMF and SMS
components on systems with a release of DFSMS/MVS before DFSMS/MVS V1R4
installed to prevent failures when the construct is accessed. The four data class
attributes cannot be displayed, listed, defined, or altered through ISMF on a
down-level release.

The BWO, LOG, and LOGSTREAM ID attributes are saved in the RLS cell in the
catalog. The RLS toleration PTFs will fail OPEN on releases before DFSMS/MVS
V1R3 if the RLS cell exists for the VSAM sphere.

With DFSMS/MVS V1R4 most VSAM data sets can be defined completely through
JCL without requiring a separate IDCAMS DEFINE or IDCAMS ALTER.

3.6 DCOLLECT Enhancements

DCOLLECT has been modified to provide the volser number from which the
source is obtained for migrate or backup operations. The volser is the first
original volume of a multivolume data set.

A new parameter for the DCOLLECT command enables you not to produce
output for unneeded volumes. The new parameter, EXCLUDEVOLUMES, enables
you to eliminate volumes identified by the STORAGEGROUP and VOLUMES
parameters, which enable you to specify a set of volumes for which data is to be
collected.

Chapter 3. DFSMSdfp VSAM 47

 In addition, changes have been made to reporting the VSAM association
information and the actual device type information. The device type shown in
reports created on a system before DFSMS/MVS V1R4 is the logical device type
for the volume (3380, 3390).

3.6.1 DCOLLECT OUTPUT

New fields are added to the output records produced by DCOLLECT.

3.6.1.1 New Volume Field

The UMFRVOL field is a new field for migrated data (UMBFRVOL for data that is
backed up) that contains the first DASD volume as the source of migrated data
or backup data. The new field is available in the following records:
• Migrated data set information (record type M)
• Backup data set information (record type B)

Note: The size of both records (M and B) has increased by six characters.

3.6.1.2 VSAM Association Information

Record type A has been expanded by two new fields. The first record type A for
a data set is reported; this is the record for the primary, not the secondary,
volume. The new fields are:
• DCACISZ, number of bytes in a CI
• DCACICA, number of CIs in a CA

3.6.1.3 Volume Information

The DCVDPTYP field in record type V reports the true device type rather than the
generic 3380/3390. The true device type is the actual physical device type (for
example, RAMAC), as opposed to the logical device type.

3.6.1.4 Data Class Information

For data class information, record type DC has been expanded by a new field as
well as by a field extension:
• DDCRABS, the new bit - Record Access Bias is specified
• DDCRBIAS, the new field - VSAM Record Access Bias constant
− DDCRABUS, value 0 = user
− DDCRABSY, value 1 = system

3.6.2 DCOLLECT Command

With this DCOLLECT enhancement, you can restrict the amount of DCOLLECT
output that is generated for the DCOLLECT report.

The DCOLLECT command has been enhanced with the optional parameter,
EXCLUDEVOLUME. The EXCLUDEVOLUME parameter can be abbreviated as
EXCLVOL, EXV, or EXCLUDE.

Figure 8 on page 49 shows the syntax of DCOLLECT with EXCLUDEVOLUMES.

48 DFSMS/MVS V1R4 Technical Guide

 DCOLLECT ──┬─────────────────────────────────────────────────────────┬──
│ │
└─ VOLUME ──┬──────────────────────────────────────────┬──┘
│ │
├─ EXCLUDEVOLUME(volser,volser,...) ─″── ──┤
├─ STORAGEGROUP(sgname sgname,...) ────────┤
├─ MIGRATEDATA ────────────────────────────┤
├─ BACKUPDATA ─────────────────────────────┤
├─ CAPPLANDATA ────────────────────────────┤
└─ SMSDATA(SCDSNAME(entryname) | ACTIVE) ──┘

Figure 8. DCOLLECT Command Syntax

DCOLLECT can be abbreviated by DCOL. The options for EXCLUDEVOLUME are:

• A fully qualified volser, counting one to six characters
• A partially specified volser using trailing asterisks as a place holder for all
remaining characters
• Any combination of the above

You can exclude information on a selected volume or group of volumes.

Chapter 3. DFSMSdfp VSAM 49

50 DFSMS/MVS V1R4 Technical Guide
Chapter 4. DFSMShsm

DFSMShsm V1R4 enhances product useability as well as performance and

throughput. In the disaster recovery and backup area, it provides more
parallelism and better use of tape media. Therefore the DFSMShsm aggregate
backup and recovery support (ABARS) function will need fewer tape cartridges
than before to store the same amount of data. DFSMShsm V1R4 can exploit the
Magstar technology. DFSMShsm can access its VSAM control data sets (CDSs)
in RLS mode by using the coupling facility. This enhancement relieves the
parallel sysplex constraint that DFSMShsm has today and enables you to take
advantage of parallel sysplex benefits.

DFSMShsm V1R4 has been enhanced to write data to two tapes simultaneously.
Thus you can make a backup copy or migrate data to tape on two different
devices at the same time. With this new function, you can address a second,
perhaps remote, Automated Tape Library (ATL).

DFSMShsm can now communicate with DFSMSrmm and the DFSMS/MVS

Optimizer directly without exits. It no longer requires the exits to send requests
to DFSMSrmm tape management or the DFSMS/MVS Optimizer. This frees the
ARCTVEXT exit, allowing it to be used by other tape management products.

DFSMShsm will provide more account data as well as aliases for keywords on
specified DFSMShsm commands.

DFSMShsm V1R4 provides the following set of enhancements:

• Duplex tape
• Alter without recall (IDCAMS ALTER)
• ABARS
• CDS access in RLS mode
• DFSMShsm and DFSMSrmm interface
• Dump analysis elimination (DAE)
• Serviceability support items

4.1 Duplex Tape

DFSMShsm V1R4 provides an alternative to TAPECOPY processing for backup
and migration cartridge tapes. With duplex tape DFSMShsm creates two tapes
concurrently. The original tape is intended to be kept onsite, and the alternate
tape can be taken offsite or perhaps handled on a remote ATL. The pair of tapes
will maintain the original versus alternate distinction. The original tape will have
one of the following data set names:
• prefix.HMIGTAPE.DATASET
• prefix.BACKTAPE.DATASET

The alternate tape will have one of the following data set names:
• prefix.COPY.HMIGTAPE.DATASET
• prefix.COPY.BACKTAPE.DATASET

 Copyright IBM Corp. 1997 51

 Thus the new tape can be compatible with the current tapes created by
TAPECOPY. In an SMS environment, the ACS routines can direct the alternate
tape to a different tape library, such as one at an offsite ___location. In a non-SMS
environment, the output restriction (for example, SETSYS UNIT(3490)) will be
used for both the original and the alternate tape.
Note: The alternate tape must have the same tape geometry as the original, for
example, both must be 3490 Extended Capacity Cartridge Storage Tape (ECCST)
tapes.

4.1.1 Automatic TAPECOPY Scheduling

DFSMShsm automatically schedules a TAPECOPY when necessary to ensure
that valid alternate tapes exist. To this end, a new CDS record, TCN, is created
in the OCDS. Automatic TAPECOPY commands are scheduled for certain
infrequently occurring functions and exceptions. When a TAPECOPY completes
successfully, the TCN record for that volume is deleted.

This new TCN record is type E, and the key is either M- for migration tape or B-
for backup tape, followed by the 6-byte volume serial number, and padded with
blanks.

During secondary space management, if any TCN records exist for migration
volumes, another TAPECOPY is scheduled.

During autobackup on the primary host, if any TCN records exist for backup
volumes, another TAPECOPY is scheduled. This copy occurs only on the
primary host, to avoid multiple TAPECOPYs for the same volumes.

4.1.2 Supported DFSMShsm Functions

Duplex tape is supported for following DFSMShsm functions:
• Volume backup (including autobackup)
• Volume migration (including primary space management)
• Recycle
• Move backup copies from ML1 volumes
• Secondary space management
• Data set migration
• FREEVOL
− Migration volumes
− Backup volumes
− ML1BACKUPVERSIONS
• SPILL processing
• ARECOVER ML2 tape
Only one tape will be created. A TCN record will be created and written to
the CDS to be processed by secondary space management to create the
second (alternate) copy of the tape.

52 DFSMS/MVS V1R4 Technical Guide

4.1.3 Invocation
Changes have been made to the SETSYS command and the QUERY SETSYS
command in support of duplex tape.
• SETSYS command
A new keyword, DUPLEX, has been added to the SETSYS command to allow
the specification of duplexing for backup and/or migration volumes.
• QUERY SETSYS command
The QUERY SETSYS command has been enhanced to display the current
duplex status of migration and backup processing. The message is as
follows:
ARC0442I TAPE OUTPUT PROMPT FOR TAPECOPY=NO, DUPLEX
ARC0442I (CONT.) BACKUP TAPES=YES, DUPLEX MIGRATION TAPES=YES

4.1.4 Initial Tape Selection

In a duplexing environment, DFSMShsm attempts to allocate two tape drives,
mount two tapes, and write both tapes concurrently. The first tape is
predesignated as the original, and the second, as the alternate.

DFSMShsm selects tapes, using the following selection process:

• In a duplex environment
1. A partial tape and its alternate (partial tapes without alternates are
excluded), if one exists
2. If no partial tapes are found, an empty ADDVOLed volume, if one exists,
with a scratch as an alternate
3. Two scratch volumes
• In a Simplex environment:
1. A partial tape that does not have a duplicate (partial tapes with
alternates are excluded)
2. If no partial tapes are found, an empty ADDVOLed volume, if one exists.
3. A scratch volume

This selection process enables DFSMShsm to better support private scratch

pools.

4.1.5 TAPECOPY Processing

TAPECOPY continues to function as before, with three exceptions:
• When any alternate is being replaced, that is, there is an external
TAPECOPY request for its original, DFSMShsm calls the installationwide exit,
ARCTVEXT, if active, to release the previous alternate tape volume through
the tape management systems. A new flag in the ARCTVEXT parameter list
indicates that the tape being released is an alternate tape being replaced by
a user TAPECOPY; the original tape is not being released. The new flag is
also defined in the interface to RMM (EDGTVEXT).
• When the internal TAPECOPY command is successfully processed,
DFSMShsm deletes the TCN OCDS record for that volume and marks the
volume as available.

Chapter 4. DFSMShsm 53
 • When an internal TAPECOPY MWE is being processed, if a valid alternate
exists in the TTOC record, do not perform TAPECOPY but delete only the
TCN record.

4.1.6 Tape Exception Processing

The actions that DFSMShsm takes in a duplex environment when a tape
exception occurs are described below.

4.1.6.1 Allocating and Mounting an Existing Alternate

If DFSMShsm allocates and mounts an empty original tape for double output but
is unable to allocate and mount a new alternate scratch volume, it continues to
write the original volume. However, in this case, when the original volume
(ADDVOLed or scratch) is demounted, DFSMShsm marks the volume available
after a successful TAPECOPY.

4.1.6.2 I/O Error Processing

If the duplexed original tape encounters an I/O error, both tapes are marked full
and demounted and two different tapes are selected. The user data set, related
to the I/O error, is restarted from the beginning on the different tapes.

If the alternate tape encounters an I/O error, it is demounted and returned to

scratch, but the original continues to be written until it is demounted. The
original is marked as duplexed and unavailable, and DFSMShsm automatically
schedules a TAPECOPY to create an alternate tape. DFSMShsm makes the
volume available after a successful TAPECOPY.

4.1.6.3 End-of-Volume Processing

Should the original tape reach natural EOV, DFSMShsm markd it full and keeps
both tapes if they contain any valid data. If the tapes do not contain any valid
data, they are released. Regardless of whether the tapes contain valid data, the
data set being processed when the natural EOV was encountered is restarted
from its beginning on two new tapes.

If the alternate tape encounters natural EOV, it is demounted and returned to

scratch, but the original continues to be written until it is demounted. The
original is marked as duplexed and unavailable, and DFSMShsm automatically
schedules a TAPECOPY to create an alternate tape. DFSMShsm makes the
volume available after a successful TAPECOPY.

4.1.6.4 Percent-Full End-of-Volume Processing

When the original tape reaches its percent-full capacity, before performing a
FEOV, DFSMShsm must flush all data to both tapes. If natural EOV is reached on
either tape, DFSMShsm marks the tapes as FULL and restarts processing the
current data set from its beginning on two new tapes.

4.1.6.5 Exception Handling

During some exception processing, DFSMShsm automatically schedules an
internal TAPECOPY to create the alternate. See section 4.1.1, “Automatic
TAPECOPY Scheduling” on page 52 for a description of automatic TAPECOPY
processing.

54 DFSMS/MVS V1R4 Technical Guide

4.1.7 AUDIT Processing
AUDIT reports discrepancies between the MCV/MCT record and the TTOC
relating to the alternate volume. AUDIT does not process TCN records.

4.1.8 LIST Processing

The (H)LIST MVOL output includes the duplex status and the alternate volser.

The (H)LIST BVOL output includes the duplex status and the alternate volser.

4.1.9 Limitations
Some parameters of nongeneric TAPECOPY commands are not available with
the duplex tape option.

The TAPECOPY command supports a private DFSMShsm scratch pool

environment through the ALTERNATEVOLUMES and INDATASET parameters.
Using those parameters, you can specify a target tape to which a specific input
tape will be copied. With the INDATASET parameter, you can also specify a
specific expiration date for each alternate tape. With the duplex option, all
allocations are for a nonspecific tape, and all expiration dates are the same,
subject to data class capabilities.

The TAPECOPY command provides an ALTERNATEUNITNAME parameter that

can be useful in non-SMS allocations and SMS filtering. You can specify a unit
name specifically for allocation of the alternate tape. No such capability is
available when using the duplex option, so the unit name for the original and
alternate will be the same. If ACS routines are used, the tape data set name can
be used to direct the different allocations. The alternate tape data set name
inserts a COPY qualifier that is not present in the original tape data set name.

4.1.10 Migration and Coexistence

Previous releases of DFSMShsm require toleration PTFs for all functions that
cause movement of user data to tape. These functions must not attempt to
extend original duplexed tapes. Such tapes will be bypassed in the selection
process.

If changing to use duplexed output and you have been using the
PARTIALTAPE(REUSE) option, consider marking full the existing partial tapes as
they are not considered in a duplexing environment. Only by marking partial
tapes full will they be considered by recycle.

4.2 Alter without Recall

Today, if you want to alter the catalog entry from a migrated data set,
DFSMShsm performs a recall to bring back the data set on level 0. DFSMShsm
performs a recall even if you do not change the contents of the data set.

You thus waste time until the data set is recalled and incur unnecessary CPU
and I/O use. Alter without recall relieves this constraint by not recalling a
migrated data set to perform an IDCAMS ALTER entryname command for only
the storage class and/or management class submitted by an end user. Thus, an
IDCAMS ALTER entryname command for only the storage class and/or
management class submitted by an end user will cause the storage class and/or

Chapter 4. DFSMShsm 55
 management class to be altered for a migrated data set without recalling the
data set.

The DFSMShsm record used for SMS space management processing has been
updated with the new storage class and/or management class to keep it
consistent with the construct names in the catalog.

If DFSMShsm fails to process the catalog update request for the requested class
names, catalog will not update the class name(s) in the catalog and the IDCAMS
command will fail.

When catalog fails to update the class name(s) in the catalog after DFSMShsm′ s
updates, the DFSMShsm record will retain the updates made. The IDCAMS
command will fail.

The new alter without recall function causes both DFSMShsm′s and catalog′ s
records to be updated and kept synchronized. The HALTER EXEC updates only
DFSMShsm records, leaving DFSMShsm′s and catalog′s records
unsynchronized. The HALTER EXEC has been modified to reflect the fact that
IDCAMS ALTER can now perform the alter without recall function directly.

4.2.1 New IDCAMS Interface with Catalog

IDCAMS can now use the LOCATE macro without a DFSMShsm recall. The
LOCATE macro results in an SVC 26 (IGG026DU) call.

In the IDCAMS ALTER datasetname for only the STORAGECLASS and/or

MANAGEMENTCLASS path, IDCAMS calls catalog for both a LOCATE command
and an ALTER request with a do not recall the data set indicator . IGG026DU will
not recall the data set, even if it is migrated.

IGG026DU sees the command in both cases but due to the do not recall the data
set indicator being set, the request will pass directly on the catalog.

4.2.2 New Catalog Notification of DFSMShsm

When catalog receives one of the following requests from IDCAMS:
ALTER dsn STORAGECLASS(storageclassname)
MANAGEMENTCLASS(managementclassname)
ALTER dsn STORAGECLASS(storageclassname)
ALTER dsn MANAGEMENTCLASS(managementclassbame)
and it has successfully completed the security checking for the data set and
the data set is migrated, it notifies DFSMShsm to update its inventory with these
values.

4.2.3 DFSMShsm Processing Considerations

DFSMShsm updates the CDS MCD record for the data set with the new SMS
constructs sent in the ARCHSEND text string from the catalog. DFSMShsm does
not update the CDS backup records.

DFSMShsm sends a return code back to the catalog, indicating the success or
failure of the CDS record update. When the return code from DFSMShsm is not
zero, the catalog update will not be performed and the IDCAMS alter function
will fail.

56 DFSMS/MVS V1R4 Technical Guide

 If the catalog updates should fail after DFSMShsm has successfully made its
update, the DFSMShsm updates will be retained. The IDCAMS command will fail.
An IDCAMS message informs you about the out of Sync status of catalog and
DFSMShsm.

4.2.4 DFSMShsm Error Processing

If the update request sent to DFSMShsm from the catalog fails, there is no retry
logic in DFSMShsm. If DFSMShsm is stopped or shut down or its address space
is canceled while the update request from catalog is queued to DFSMShsm, the
MWE will be posted complete with a nonzero return code. DFSMShsm will not
process the request.

4.3 ABARS
The following ABARS enhancements are included in DFSMShsm V1R4:
• ABARS output files can be stacked on a minimum number of tape volumes
• Number of concurrent active ABARS requests has increased to 64.
• Invocation of ARCBEEXT has been extended to data sets that are being
dumped by DFSMSdss processing so that installations can bypass data sets
that fail.
• GDG base names can be specified in the ALLOCATE list. Thus you can back
up and restore GDG base definitions without having to back up an associated
generation data set (GDS).
• ABARS activity log on DASD or tape is automatically deleted when
aggregate roll-off occurs, during either automatic roll-off or EXPIREBV
processing.
• CPU time for processing ABACKUP or ARECOVER is maintained in the
ABACKUP/ARECOVER Function Statistics record (WWFSR) and aggregate
backup and recovery (ABR) record. A new 32-character accounting
information attribute in the aggregate group definition is saved in the
WWFSR and the ABR record.
• The TGTGDS and OPTIMIZE keywords used when invoking DFSMSdss to
dump and restore data sets are externalized.

4.3.1 Output File Stacking

Today ABARS ABACKUP typically needs at least three tape cartridges for output:
one for the control file ABARS ARECOVER needs for the recovery process, one
to contain the activity log and instruction data set, and one or more for the data
itself.

ABARS provides a new function that allows the ABACKUP output files to be
stacked on a minimum number of tape cartridges. The minimum number of tape
cartridges could be 1 if the aggregate is small.

Chapter 4. DFSMShsm 57
 4.3.1.1 Invocation of Output File Stacking
The SETSYS command has been updated with a new parameter,
ABARSTAPES(STACK|NOSTACK), with STACK as default.

The STACK parameter indicates to ABACKUP to stack the ABACKUP output files
onto a minimum number of tape volumes, NOSTACK indicates to ABACKUP not
to stack the ABARS tape. Specifying NOSTACK causes ABACKUP to operate as
it does in earlier releases of DFSMShsm.

4.3.1.2 QUERY SETSYS and QUERY ABARS

The QUERY SETSYS and QUERY ABARS commands have been enhanced to
display the status of the ABARSTAPES parameter in message ARC6036I.

4.3.1.3 ABACKUP and ARECOVER DATASETNAME

The ABACKUP command has been enhanced with the new STACK|NOSTACK
parameters to allow installations to override the SETSYS ABARSTAPES.

The ARECOVER command with the DATASETNAME parameter allows new

STACK|NOSTACK keywords to be specified. Specifying STACK|NOSTACK
indicates to ARECOVER DATASETNAME processing whether or not the
ABACKUP output was stacked. Whether output was stacked must be known
before mounting the control file tape because ARECOVER has to allocate file
sequence number 4 if STACKED or not specify a file sequence number if not
STACKED.

The STACK|NOSTACK parameter is unnecessary on the ARECOVER command

specified with the AGGREGATE parameter. ARECOVER AGGREGATE obtains all
information about ABACKUP output files from the catalog. Thus it will knows
whether or not to allocate with an appropriate file sequence number.

4.3.2 Up to 64 Concurrent Requests

ABARS has been enhanced to allow up to 64 concurrent ABARS address spaces.
The current restrictions remain; that is, you cannot simultaneously back up the
same aggregate group name or recover using the same aggregate group name
or control file data set name.

4.3.2.1 Invocation
The existing SETSYS MAXABARSADDRESSSPACE(n) parameter has been
enhanced to allow the value of n to range from 1 to 64, with the default
remaining as 1.

The started task identifier contains the task number as part of the identifier. The
identifier is currently ABARnntt, where nn is the task number, and tt is a time
stamp. The value of nn can now range from 1 to 64. For example, task 56 would
get an identifier of ABAR56tt.

4.3.3 Invocation of ARCBEEXT Extended to DFSMSdss Processing

ABACKUP processing has been enhanced to extend the error conditions where
the ARCBEEXT installationwide exit gets control. With this enhancement
ARCBEEXT gains control when DFSMSdss errors occur while dumping level 0
DASD data sets in the INCLUDE list.

The ARCBEEXT installationwide exit is now called if SETSYS EXITON(BE) is

specified and one of the following failures occurs:

58 DFSMS/MVS V1R4 Technical Guide

 • An E level DFSMSdss message is issued indicating that a data set has failed
dump processing.
• A W level DFSMSdss messages is issued and DFSMSdss does not indicate
that the warning is OK.

When one of the above failures occurs, the failing data set name and an
indicator that a DFSMSdss failure occurred are passed to ARCBEEXT. Thus you
can determine whether the data set should be skipped. If ARCBEEXT does not
indicate that the data set be skipped, ABACKUP will fail at that point; otherwise
the data set in error will be skipped and the next data set processed.

To enable this support, you must specify SETSYS EXITON(BE) before issuing the
ABACKUP command.

4.3.4 GDG Base Name in ALLOCATE Statement

Customers have asked for a method where they can have ABARS define a GDG
base at the recovery site without having to back up any of the GDSs. Older
versions of DFSMShsm do not permit the specification of a GDG base name in
the selection data set. If a GDS name was specified, however, ABARS
automatically retrieved the GDG base definition and maintained it in the control
file. This support will remain in effect.

ABARS has been enhanced so that you can specify a GDG base name in the
ALLOCATE statement and have ABARS predefine the GDG base name during
ARECOVER processing if it does not currently exist. You can specify a GDG
base name even though you have not specified an associated GDS name in the
selection data set.

During ARECOVER processing, all GDG base names are defined before any
GDSs specified in the aggregate are restored.

4.3.5 Automatic Delete of ABARS Activity Log during Roll-Off Processing

A new SETSYS parameter, ABARSDELETEACTIVITY(Y | N), enables you to select
whether or not you want to have DFSMShsm automatically delete the ABARS
activity logs during ABARS roll-off processing or EXPIREBV ABARSVERSIONS
processing.

If you select ABARSDELETEACTIVITY(Y), the ABARS activity log is deleted during

either automatic roll-off of expired ABARS versions or EXPIREBV
ABARSVERSIONS processing.

If you specify ABARSDELETEACTIVITY(N), there is no automatic deletion.

The default is ABARSDELETEACTIVITY(N).

4.3.6 CPU Time for Aggregate Processing in WWFSR

ABARS now maintains the CPU time for processing ABACKUP and ARECOVER
requests in the WWFSR control block. A WWFSR is created at the completion of
each individual ABACKUP and ARECOVER request and can be written as an
SMF record if requested.

The ABR record has been enhanced to maintain an ABACKUP CPU time and an
ARECOVER CPU time.

Chapter 4. DFSMShsm 59
 If an ARECOVER request fails and is reissued with a valid RESTART data set, the
CPU time in the WWFSR reflects only the time to process the remaining data
sets. The ABR record, however, accumulates the CPU times of each restart until
the recovery of the aggregate is successful.

The ISMF aggregate group definition panels have been enhanced to allow
specification of a 32-character accounting code. This accounting code is also
written to the WWFSR control block, ABR record, and the ABACKUP control file.
It is written to the control file so that the account code can be propagated to the
recovery site without an aggregate definition at the recovery site.

If you specified SETSYS SMF(smfid), ABARS writes the WWFSR as an SMF

record with the record number (smfid+1) in the SYS1.MANx or SYS1.MANy
system data set. You can then use the reported CPU time to assist in
chargeback of ABARS requests.

4.3.7 TGTGDS and OPTIMIZE Keyword Externalized

The TGTGDS and OPTIMIZE keywords that ABARS passes to DFSMSdss to dump
and restore data sets are externalized.

4.3.7.1 TGTGDS Keyword

The TGTGDS keyword that ARECOVER processing passes to DFSMSdss during
restore processing is externalized in this release of DFSMShsm. This provides
greater flexibility managing SMS-managed GDSs that are being restored to level
0 DASD.

ABARS introduces a new SETSYS ARECOVERTGTGDS(option) parameter.

Valid values for option are:

• DEFERRED
Specifies that the target data set is to be assigned the DEFERRED status
• ACTIVE
Specifies that the target data set is to be assigned the ACTIVE status, for
example, rolled into the GDG base
• ROLLEDOFF
Specifies that the target data set is to be assigned the ROLLEDOFF status
• SOURCE
Specifies that the target data set is to be assigned the same status as that of
the source data set. This is the default; it maintains consistency with
releases of DFSMShsm previous to DFSMShsm V1R4.

The installation can override the SETSYS ARECOVERTGTGDS with a new

keyword on the ARECOVER command, TGTGDS(option). The values for option
are the same as for ARECOVERTGTGDS.

60 DFSMS/MVS V1R4 Technical Guide

 4.3.7.2 OPTIMIZE Keyword
The OPTIMIZE keyword that ABACKUP processing passes to DFSMSdss during
dump processing of level 0 data sets is externalized in this release of
DFSMShsm. This keyword enables installations to adjust performance when
backing up level 0 DASD data sets that are specified in the INCLUDE list.

ABARS introduces a new SETSYS ABARSOPTIMIZE(n) parameter. The value

specified for n controls what ABACKUP will specify in the OPTIMIZE keyword
when invoking DFSMSdss to back up level 0 DASD data sets.

Valid values for n are as follows:

• 1 DFSMSdss reads one track at a time.
• 2 DFSMSdss reads two tracks at a time.
• 3 DFSMSdss reads three tracks at a time.
• 4 DFSMSdss reads a cylinder at a time.

The default for n is 3.

The installation can override the specification for n in the SETSYS

ABARSOPTIMIZE keyword by specifying a new OPTIMIZE(option) keyword on the
ABACKUP command. The values for option are the same as for
ABARSOPTIMIZE.

If ABACKUP OPTIMIZE is not specified, ABACKUP will use the value defined for
n in the SETSYS ABARSOPTIMIZE(n) keyword.

4.3.8 ISMF Changes for ABARS Accounting Information

The following ISMF Aggregate Group Applications are affected by these ISMF
changes:
• Aggregate Group Define/Alter
• Aggregate Group Display
• Aggregate Group List Display
• Aggregate Group List Sort
• Aggregate Group List View
• Aggregate Group List Print

4.3.9 Migration and Coexistence

There are migration and coexistence concerns if you have multiple systems with
both DFSMShsm V1R4 and earlier releases of DFSMShsm installed.

4.3.9.1 Disaster Recovery Site Considerations

If you specify the NOSTACK parameter on the ABACKUP command when
backing up on a DFSMShsm V1R4 system, you can successfully ARECOVER on a
downlevel DFSMShsm.

Chapter 4. DFSMShsm 61
 4.3.9.2 Migration
Use the NOSTACK parameter on the ARECOVER DATASETNAME command to
ARECOVER backups performed on a downlevel version of DFSMShsm.

4.3.9.3 Coexistence
Use the ARECOVER AGGREGATE command on a DFSMShsm V1R4 system to
ARECOVER a backup performed on a downlevel DFSMShsm. This command
also allows a downlevel version of DFSMShsm to ARECOVER an aggregate
backed up on DFSMShsm V1R4, whether or not the output files are stacked.

4.4 CDS Record Level Sharing

DFSMShsm suffers data set contention against its CDSs when three or more
hosts are processing in a shared environment. This is especially true when a
work profile has a large amount of small data sets for DFSMShsm to process.
DFSMShsm V1R4 accesses its CDSs in VSAM RLS mode, which relieves this
contention and improves throughput and performance of DFSMShsm functions.

DFSMShsm V1R4 provides an option for its CDSs to be accessed in RLS mode.
Thus DFSMShsm can take advantage of the S/390 coupling facility to reduce CDS
contention.

For a brief description of the benefits of a Parallel Sysplex with a coupling facility
and how it works, see Appendix A, “Parallel Sysplex and VSAM Record Level
Sharing” on page 139, which also contains an overview of VSAM RLS and VSAM
RLS locking.

DFSMShsm uses VSAM RLS with LOG(NONE), which means that DFSMShsm
does not use the MVS Logger facility. The DFSMShsm VSAM RLS
implementation uses the CACHE and LOCK coupling facility structures. The
DFSMShsm CDS recover process is unchanged. DFSMShsm still uses its own
journal for forward recovery.

4.4.1 Invocation
Startup procedure keyword CDSSHR has been enhanced to accept RLS as a
parameter. When RLS is specified, the DFSMShsm CDSs are accessed in RLS
mode. When the CDSs are accessed in RLS mode, any values specified for
CDSQ and CDSR are ignored.

CDSSHR = {YES|RLS|NO}
YES Perform multiple-processor serialization of the type requested
by the CDSQ and SDSR keywords.

RLS Perform multiple-processor serialization, using RLS.

NO Do not perform mulitple-processor serialization.

If the CDSSHR keyword is not specified, multiple-processor serialization

(non-RLS mode) is used if the index component of the MCDS resides on a DASD
volume that has been SYSGENed as SHARED or SHAREUP. This process
remains unchanged from the previous release.

62 DFSMS/MVS V1R4 Technical Guide

4.4.2 CDS Access in Record Level Sharing Mode
When DFSMShsm is active, user programs and ISV products that access the
CDSs must use the same access technique as that DFSMShsm currently uses.

4.4.3 QUERY CONTROLDATASETS Command

The QUERY CONTROLDATASETS command has been enhanced to display the
CDS serialization technique in use and the boundaries being used for dynamic
key boundary multiclusters.

4.4.4 Multicluster CDSs

When RLS is used to access the CDSs, they cannot be defined with key ranges.
The multicluster support for the CDS has been enhanced so that the
multiclusters can be defined without key ranges. DFSMShsm dynamically
calculates the key boundaries used for each cluster.

DFSMShsm continues to support key ranges for CDSs not accessed in RLS
mode.

4.4.5 CDS Creation or Redefinition

When you use dynamic key boundary multicluster support, the CDSs cannot be
empty when you start DFSMShsm.

4.4.5.1 Sample JCL for Multicluster, Non-Key-Range, and

RLS-Eligible CDSs
The following is sample JCL for redefining the CDSs as multicluster,
non-key-range, and RLS eligible. This sample JCL is shipped in ARCTOOLS.
//HSMCDS JCL ,MSGLEVEL=(1,1)
//*******************************************************************/
//* SAMPLE JCL THAT ALLOCATES NEW MULTICLUSTER, NONKEY-RANGE, RLS */
//* ELIGIBLE CONTROL DATA SETS AND COPIES THE OLD KEY-RANGE CONTROL */
//* DATA SET INTO THEM */
//*******************************************************************/
//*
//STEP1 EXEC PGM=IDCAMS,REGION=512K
//SYSPRINT DD SYSOUT=A
//SYSUDUMP DD SYSOUT=A
//SYSIN DD *
DELETE DFHSM.RLS.MCDS1 CLUSTER
DELETE DFHSM.RLS.MCDS2 CLUSTER
DELETE DFHSM.RLS.BCDS1 CLUSTER
DELETE DFHSM.RLS.BCDS2 CLUSTER
DELETE DFHSM.RLS.OCDS CLUSTER

DEFINE CLUSTER (NAME(DFHSM.RLS.MCDS1) -

STORAGECLASS(STRCLRLS) -
CYLINDERS(2) NOIMBED NOREPLICATE -
RECORDSIZE(200 2040) FREESPACE(0 0) -
INDEXED KEYS(44 0) SHAREOPTIONS(3 3) -
UNIQUE LOG(NONE)) -
DATA -
(NAME(DFHSM.RLS.MCDS1.DATA) -
CONTROLINTERVALSIZE(12288)) -
INDEX -
(NAME(DFHSM.RLS.MCDS1.INDEX) -

Chapter 4. DFSMShsm 63
 CONTROLINTERVALSIZE(4096))

DEFINE CLUSTER (NAME(DFHSM.RLS.MCDS2) -

STORAGECLASS(STRCLRLS) -
CYLINDERS(2) NOIMBED NOREPLICATE -
RECORDSIZE(200 2040) FREESPACE(0 0) -
INDEXED KEYS(44 0) SHAREOPTIONS(3 3) -
UNIQUE LOG(NONE)) -
DATA -
(NAME(DFHSM.RLS.MCDS2.DATA) -
CONTROLINTERVALSIZE(12288)) -
INDEX -
(NAME(DFHSM.RLS.MCDS2.INDEX) -
CONTROLINTERVALSIZE(4096))

DEFINE CLUSTER (NAME(DFHSM.RLS.BCDS1) -

STORAGECLASS(STRCLRLS) -
CYLINDERS(2) NOIMBED NOREPLICATE -
RECORDSIZE(200 2040) FREESPACE(0 0) -
INDEXED KEYS(44 0) SHAREOPTIONS(3 3) -
UNIQUE LOG(NONE)) -
DATA -
(NAME(DFHSM.RLS.BCDS1.DATA) -
CONTROLINTERVALSIZE(12288)) -
INDEX -
(NAME(DFHSM.RLS.BCDS1.INDEX) -
CONTROLINTERVALSIZE(4096))

DEFINE CLUSTER (NAME(DFHSM.RLS.BCDS2) -

STORAGECLASS(STRCLRLS) -
CYLINDERS(2) NOIMBED NOREPLICATE -
RECORDSIZE(200 2040) FREESPACE(0 0) -
INDEXED KEYS(44 0) SHAREOPTIONS(3 3) -
UNIQUE LOG(NONE)) -
DATA -
(NAME(DFHSM.RLS.BCDS2.DATA) -
CONTROLINTERVALSIZE(12288)) -
INDEX -
(NAME(DFHSM.RLS.BCDS2.INDEX) -
CONTROLINTERVALSIZE(4096))

DEFINE CLUSTER (NAME(DFHSM.RLS.OCDS) -

STORAGECLASS(STRCLRLS) -
CYLINDERS(2) NOIMBED NOREPLICATE -
RECORDSIZE(200 2040) FREESPACE(0 0) -
INDEXED KEYS(44 0) SHAREOPTIONS(3 3) -
UNIQUE LOG(NONE)) -
DATA -
(NAME(DFHSM.RLS.OCDS.DATA) -
CONTROLINTERVALSIZE(12288)) -
INDEX -
(NAME(DFHSM.RLS.OCDS.INDEX) -
CONTROLINTERVALSIZE(4096))
/*
//*******************************************************************/
//* COPY THE OLD CONTROL DATA SETS INTO THE NEWLY DEFINED */
//* MULTICLUSTER CONTROL DATA SETS */
//* NOTE: THE FROMKEY/TOKEY VALUES ARE ONLY SAMPLES. THE ACTUAL */
//* PARAMETERS USED FOR THESE KEYWORDS SHOULD BE DERIVED FROM */

64 DFSMS/MVS V1R4 Technical Guide

 //* ACTUAL CDSS BEING USED. */
//*******************************************************************/
//*
//STEP2 EXEC PGM=IDCAMS,REGION=512K
//SYSPRINT DD SYSOUT=A
//SYSUDUMP DD SYSOUT=A
//SYSIN DD *
REPRO INDATASET(DFHSM.MCDS) OUTDATASET(DFHSM.RLS.MCDS1) -
FROMKEY(X′00′) TOKEY(DFHSM)
REPRO INDATASET(DFHSM.MCDS) OUTDATASET(DFHSM.RLS.MCDS2) -
FROMKEY(DFHSN) TOKEY(DFHSM)
REPRO INDATASET(DFHSM.BCDS) OUTDATASET(DFHSM.RLS.BCDS1) -
FROMKEY(X′00′) TOKEY(DFHSM)
REPRO INDATASET(DFHSM.BCDS) OUTDATASET(DFHSM.RLS.BCDS2) -
FROMKEY(DFHSN) TOKEY(DFHSM)
REPRO INDATASET(DFHSM.OCDS) OUTDATASET(DFHSM.RLS.OCDS) -
/*

4.4.6 DCOLLECT
DCOLLECT has been changed so that it first attempts to open the CDSs in RLS
mode. If the CDSs are not RLS eligible, OPEN error message IDC161I RC9 is
issued. This message can be ignored for non-RLS-eligible data sets because the
OPEN is retried for non-RLS mode. If both OPENs fail, a new DFSMShsm
message is issued.

4.4.7 ARCIMPRT
The parameter passed to ARCIMPRT, the enhanced CDS recovery function utility,
specifies which cluster of a multicluster CDS should be recovered.

If you are recovering a single-cluster CDS, the parameters are:

BCDS | BACKUPCONTROLDATASET
MCDS | MIGRATIONCONRTOLDATASET
OCDS | OFFLINECONTROLDATASET

If you are recovering a single cluster of a multicluster CDS, the parameters are:
BCDS n | MCDSCONTROLDATASET n
MCDS n | MIGRATIONCONRTOLDATASET n
where n is a value from 1 to 4. (The OCDS cannot be defined as a
multicluster.)

4.4.8 CDS Backup

When DFSMShsm is active and accessing CDSs in RLS mode, DFSMSdss must
be specified as the datamover for CDS backup. If directed to tape, the
PARALLEL parameter must also be specified. If either of these conditions is not
met during automatic CDS version backup, these values will override the
existing values, and an informational message will be issued. If either of these
conditions is not met when BACKVOL CDS is issued, the command will fail.

Chapter 4. DFSMShsm 65
4.4.9 Migration
Changes to the DFSMShsm CDSs and the MVS/ESA or OS/390 systems are
required before the CDSs can be accessed in RLS mode.

4.4.9.1 Required CDS Changes

Before the CDSs can be accessed in RLS mode, they must be defined or altered
to be RLS eligible. The following required changes must be made to all three
CDSs (BCDS, MCDS, OCDS), or DFSMShsm will fail at startup. All three CDSs
are accessed in the same mode, either RLS or non-RLS.
• The CDSs must be defined or altered with the LOG(NONE) attribute.
Example: ALTER cdsname LOG(NONE)
Do not use the ALL or UNDO parameter of LOG.
To change the CDSs back to non-RLS eligible, use the ALTER cdsname
NULLIFY(LOG) command.
• The CDSs must be SMS managed to use RLS. The SMS configuration must
include a storage class definition that indicates which coupling facility to use.
• All operating systems running with DFSMShsm must be coupling facility
capable (MVS/ESA SP Version 5.2 and up), and the processors must have
access to the coupling facility.
• The CDSs cannot be defined as key-range data sets. If the CDSs were
previously defined as multicluster, they used key ranges and must be
redefined without key ranges. DFSMShsm dynamically calculates the key
boundaries used for the redefined CDSs.

4.4.9.2 Required CDS Backup and Datamover Changes

If CDS backup will be invoked while the CDSs are accessed in RLS mode,
DFSMSdss must be the specified datamover. If the backup is directed to tape,
the PARALLEL parameter of SETSYS CDSVERSIONBACKUP(BACKUPDEV
ICECATEGORY) must be specified.

4.4.10 Coexistence
When one system accesses the DFSMShsm CDSs in RLS mode, all systems in
the hsmplex must access the CDSs in RLS mode.

4.4.10.1 Processing Concurrently at DFSMShsm V1R4 Level

DFSMShsm fails at startup if the selected serialization technique of the starting
DFSMShsm differs from the serialization technique of other DFSMShsm systems
already active in the hsmplex.

4.4.10.2 Processing Concurrently with a Downlevel Version of

DFSMShsm
DFSMShsm fails at startup on a downlevel system if an active DFSMShsm ViR4
system is accessing the CDSs in RLS mode. A toleration PTF must be installed
for this support.

DFSMShsm fails at startup on a downlevel system if the CDSs are defined by

using multicluster dynamic key boundaries. No toleration PTF is needed for this,
as the downlevel system determines that the CDSs are invalid and fails startup
with the ARC0900I RC8 message.

66 DFSMS/MVS V1R4 Technical Guide

4.5 DFSMShsm and DFSMSrmm Interface
With DFSMS/MVS V1R4, DFSMShsm can now communicate with DFSMSrmm
directly without exits. It no longer requires the exits to send requests to
DFSMSrmm tape management. Thus the ARCTVEXT exit is free and other tape
management products can use it.

4.5.1 Current Processing

DFSMSrmm ships source code and object code for ARCTVEXT. The
DFSMSrmm-supplied code links to the DFSMSrmm programming interface,
EDGDFHSM, to notify DFSMSrmm of tapes being released from DFSMShsm.

EDGDFHSM checks whether DFSMSrmm is in use and issues the EDG8021E

message warning that DFSMSrmm code is in use on the system when it should
not be. If DFSMSrmm should be active, it issues a WTOR, EDG8011D, to make
DFSMSrmm active.

4.5.2 New Processing

DFSMSrmm no longer ships source code and object code for ARCTVEXT.
Instead, a new general-use programming interface, EDGTVEXT, is provided,
which is called from DFSMShsm with the same parameter list as ARCTVEXT.
The EDGTVEXT is basically the same code as the old DFSMSrmm-supplied
ARCTVEXT but is now OCO and must use DFSMSrmm macros and check
whether DFSMSrmm is in use. If DFSMSrmm is not in use, it no longer calls
EDGDFHSM.

DFSMShsm always invokes EDGTVEXT before it determines whether the call to

ARCTVEXT is to be made. The ARCTVEXT exit is still invoked based on the
SETSYS EXITON() / EXITOFF() command. The EDGTVEXT general-use
programming interface frees ARCTVEXT from DFSMSrmm usage, so you can use
ARCTVEXT for other tape management systems.

4.5.3 Running DFSMShsm with DFSMSrmm

You do not use the ARCTVEXT installationwide exit to communicate between
DFSMShsm and DFSMSrmm. DFSMShsm automatically calls DFSMSrmm to
process tapes that are to be returned to the DFSMShsm tape pool or deleted
from DFSMShsm. If DFSMSrmm is not in use, the DFSMSrmm code returns to
DFSMShsm taking no action.

4.5.4 EDGDFHSM Program Interface

DFSMSrmm uses the EDGDFHSM program interface to release DFSMShsm tape
volumes, but you do not have to take any action to activate or call it. The
interface between DFSMShsm and DFSMSrmm is automatically in place when
you install DFSMS/MVS V1R4.

4.5.5 EDGTVEXT
DFSMSrmm does not use the DFSMShsm tape volume exit, ARCTVEXT, to
manage tapes that DFSMShsm uses. It has its own interface from DFSMShsm to
avoid conflicts over use of the ARCTVEXT installationwide exit.

If you have a product with similar requirements for releasing tapes from
DFSMSrmm as DFSMShsm, you can use either the EDGDFHSM or EDGTVEXT

Chapter 4. DFSMShsm 67
 program interface. The difference between EDGDFHSM and EDGTVEXT is that
EDGTVEXT accepts the ARCTVEXT parameter list, and the EDGDFHSM interface
accepts only the first volume in the ARCTVEXT parameter list.

EDGTVEXT is a callable programming interface to DFSMSrmm that is used from

DFSMShsm, or you can use it from any other APF-authorized program that needs
to obtain the same service as the DFSMShsm ARCTVEXT exit. To run
DFSMShsm with DFSMSrmm, DFSMShsm must be defined to RACF. The
DFSMShsm user ID should be other than the default user ID and should be
defined in the started procedures table, ICHRIN03, or with the RACF 2.1
STARTED class.

4.5.5.1 Invocation
EDGTVEXT can be invoked from either the LOAD, CALL, or LINK macro.

4.5.5.2 Input
The input is a parameter list that describes the volumes DFSMShsm is releasing
and the actions required. The parameter list is identical to the list DFSMShsm
passes to ARCTVEXT.

On entry, register 1 contains a pointer to the ARCTVEXT parameter list.

On entry, register 13 contains the address of a standard 18-word save area, and
register 14 contains the return address.

4.5.5.3 Output
EDGTVEXT and EDGDFHSM issue messages when problems are encountered.
EDGTVEXT always sets the ARCTVEXT return code value to zero. It does not
always pass a zero register 15 return code back to the caller. A nonzero return
code is the register 15 return code from the subsystem request attempted by
EDGDFHSM. You can obtain information about the return codes in MVS/ESA SP
Version 5 Using the Subsystem Interface .

EDGTVEXT ensures that DFSMSrmm is in use on your system before continuing

with DFSMSrmm processing. If DFSMSrmm is not in use, it sets return code
zero and returns to its caller. If DFSMSrmm is in use or should be in use, it calls
EDGDFHSM to process the volumes passed to it in the ARCTVEXT parameter list.

4.5.5.4 Environment
EDGTVEXT must be link edited in an APF-authorized library. It runs in
AMODE(31) RMODE(ANY).

4.5.5.5 Migration
There are migration concerns to consider if you invoked ARCTVEXT for
DFSMSrmm support. The DFSMSrmm code that was previously shipped in
ARCTVEXT exit must be removed, or DFSMSrmm could be invoked twice by
DFSMShsm to remove the tape volume from its inventory. If the ARCTVEXT exit
was only invoked with the previously shipped DFSMSrmm ARCTVEXT code, turn
off the ARCTVEXT exit with the SETSYS EXITOFF(TV) command. ADSM
customers who previously invoked the ARCTVEXT exit to use DFSMSrmm
services should now invoke the new DFSMSrmm general-use programming
interface, EDGTVEXT, directly from the ADSM installationwide deletion exit.
Change the DELETIONEXIT option from ARCTVEXT to EDGTVEXT in the MVS
server options file. For more information about the ADSM deletion exit, refer to
the ADSTAR Distributed Storage Manager for MVS/ESA Administrator ′ s Guide .

68 DFSMS/MVS V1R4 Technical Guide

4.6 Dump Analysis Elimination
Dump analysis elimination (DAE) is an MVS function that eliminates duplicate
dumps in MVS systems or across MVS systems.

DFSMShsm generates dumps to gather first-failure diagnostic information.

Problems that occur multiple times, perhaps on different hosts, typically
generate a storage dump. The initial dump is helpful for diagnosing the
problem; additional dumps for the same problem usually are not needed.

Controlling DFSMShsm storage dumps is a major issue with customers; the

ability to properly suppress duplicate dumps is an important step toward
successfully managing dump data sets. DFSMShsm supplies additional
information to the DAE function to build a symptom string to be used to suppress
duplicate dumps for functions that run in the same or different systems.

4.6.1 Setup Requirements

The setup requirements are:
• SETSYS SYS1DUMP must be used. (SYS1DUMP is the DFSMShsm default.)
• SYS1.PARMLIB member ADYSETxx must be coded with the keyword
SUPPRESSALL , for example:
DAE=START,RECORDS(400),
SYSMDUMP(MATCH,UPDATE),
SVCDUMP(MATCH,UPDATE,SUPPRESSALL)
• MVS/ESA SP V4.3 introduced the ability to use DAE in a multihost
environment, thereby allowing a single DAE data set to be shared across
systems in a sysplex. The coupling services of the cross system coupling
facility (XCF) and GRS must be enabled for the DAE data set to be shared in
a sysplex environment and for dumps to be suppressed across MVS
systems.

4.6.2 Programming Systems

The applicable product release level is DFSMShsm V1R4.

Only those hosts with DFSMShsm V1R4 will use the DAE suppression for dumps
for the DFSMShsm product.

DFSMShsm V1R4 uses DAE for functions that run in both the primary address
space and the ABARS secondary address space.

4.6.3 Dumps Not Eligible

DAE does not suppress SYSABEND, SYSUDUMP, SYSMDUMP, or SNAP dumps
or dumps that originate from SLIP or DUMP operator commands. Because these
dumps are taken only on demand, suppression is not desirable.

This support does not apply to dumps produced by DFSMShsm as a result of the
TRAP command.

Chapter 4. DFSMShsm 69
4.6.4 DFSMShsm Processing in the Primary Space
For dumps taken in the primary address space that are eligible, DFSMShsm will
fill in additional fields in the SDWA to build a symptom string before taking a
dump. The following variables are filled in by DFSMShsm:
• Load module name
• Csect name
• Name of the recovery routine
• DFSMShsm component identifier
• DFSMShsm component base number (prefix)

The DAE function uses the symptom string to decide whether the dump
requested is a duplicate dump.

4.6.5 DFSMShsm Processing in the ABARS Secondary Address Space

For dumps taken in the secondary address space that are eligible, DFSMShsm
uses the SYMREC parameter on the SDUMP macro that requested the dump to
build a symptom string. The parameter contain the following information:
• Load module name for the secondary address space
• Abend code
• Csect name of failing module
• Offset in failing module
• Name of the recovery routine for the secondary address space
• DFSMShsm component identifier
• DFSMShsm component base number (prefix)
The DAE function uses the symptom string to decide whether the dump
requested is a duplicate dump.

4.7 Serviceability Support Items

DFSMShsm V1R4 includes several seviceability items. They provide improved
reporting ability and facilitate DFSMShsm command use.

4.7.1 Four New SMF Functional Statistical Record Types

Four new SMF FSR subtypes have been created for the following events:
• Type 17 - Expire data set (from L0, ML1, and ML2)
• Type 18 - Partial release data set on L0 DASD
• Type 19 - Expire incremental backup version
• Type 20 - Delete incremental backup version

Two FSRs are now a subtype 17 instead of a subtype 6. Thus record subtype 6
can be written only for user-requested migrated data set deletes. FSRs are
created when the scheduled delete of the migrated data set takes place. An
ARC0734I message is issued when the delete is scheduled for:
• Expire data set from ML1
• Expire data set from ML2

70 DFSMS/MVS V1R4 Technical Guide

 There is no change from the current processing.

Two new bits have been added to one of the flag bytes in the FSR mapping to
indicate whether the data set that is being expired is from ML1 or ML2. If
neither of these bits is set, the data set being expired is from L0. Thus you can
map the expired data set FSR subtype 17 to the different volume categories of
L0, ML1, or ML2.

Three new bits have been added to the flag bytes in the FSR mapping to
indicate:
• The data set is being deleted by its expiration data in the catalog or by the
management class attributes
• Whether the incremental backup version is being deleted by the EXPIREBV
command
• Whether the incremental backup version being deleted is on a tape volume

There may be migration concerns for the conversion of two FSRs (expire from
ML1 and expire from ML2) from a subtype 6 to a subtype 17. If the SMF subtype
6 information was previously collected, the values will change because the
expires from ML1 and ML2 are no longer included. To continue to collect the
information for expires from ML1 and ML2, subtype 17 must be collected.

4.7.2 Aliases for DFSMShsm Keywords

DFSMShsm V1R4 introduces four short aliases for long keywords. This
enhancement will simplify and shorten commands.

The following changes have been made:

• BACKUP can be used as an alias for BACKUPCONTROLDATASET on the
LIST and the HLIST commands. The BCDS alias that exists today will be
kept.
For example, the command
LIST LEVEL(qualifier) BACKUP
will give you the backup version of all data sets with the initial characters of
this qualifier . A command using the DSNAME option instead of the LEVEL
option also applies:
LIST DSNAME(datasetname) BACKUP
• MIGRAT can be used as an alias for the MIGRATIONCONTROLDATASET on
the LIST and HLIST commands. The MCDS alias that exists today is still
valid. The LIST LEVEL and the LIST DSNAME commands are also applicable
to the MIGRAT alias.
• The abbreviations UUT and NOUUT can be used, respectively, for
USERUNITTABLE and NOUSERUNITTABLE on the SETSYS command.

4.7.3 AUDIT DATASETCONTROLS

The granularity of the AUDIT DATASETCONTROLS restart (RESUME) function has
been improved by saving all 44 characters of the data set name.

Chapter 4. DFSMShsm 71
4.7.4 DFSMS/MVS Optimizer HSM Monitor/Tuner
Currently DFSMShsm communicates with the Optimizer Monitor/Tuner through
the DFSMShsm initialization exit ARCINEXT. In DFSMShsm V1R4, DFSMShsm
directly invokes GFTEMINX. It utilizes a new interface, GFTEMSDX, to the
DFSMS/MVS Optimizer component.

DFSMShsm always invokes GFTEMINX during startup. If the invocation fails,

DFSMShsm continues without any error messages. The ARCINEXT exit is still
invoked based on the SETSYS EXITON() / EXITOFF() command. Thus the
ARCINEXT is freed from Optimizer use, so you can use ARCINEXT for other
reasons.

If the invocation of the GFTEMINX is successful, GFTEMSDX is always be invoked

during DFSMShsm shutdown. If the GFTEMSDX invocation fails, DFSMShsm
continues without any error messages.

There are migration concerns to consider if ARCINEXT was previously invoked

for Optimizer support. The code that invokes the GFTEMINX exit must be
removed from the ARCINEXT installationwide exit; otherwise the Optimizer
GFTEMINX exit could be invoked twice during startup. If the ARCINEXT exit was
invoked only for the Optimizer GFTEMINX exit, turn off the ARCINEXT exit with
the SETSYS EXITOFF(IN) command.

72 DFSMS/MVS V1R4 Technical Guide

Chapter 5. DFSMSdss Multivolume Selection Enhancements

The DFSMSdss multivolume selection changes affect logical data set DUMP, data
set COPY, and CONVERTV processing. The support adds a new selection
keyword, SELECTMULTI(ALL|ANY|FIRST). This keyword replaces and improves
on the current ALLMULTI keyword, which will become obsolete.

For logical data set dump, data set copy, and CONVERTV processing, the new
keyword enables a multivolume data set to be selected according to whether the
volume list includes all of the volumes on which the data set resides (ALL),
some of the volumes (ANY), or at least the volume that contains the first extent
(FIRST).

Current DFSMSdss jobs that use the ALLMULTI keyword will continue to run as
before with the specification of ALLMULTI mapping to SELECTMULTI(ANY),
unless CONVERTV to SMS is specified. In this case ALLMULTI will map to
SELECTMULTI(FIRST). If neither ALLMULTI nor SELECTMULTI is specified, the
default will be SELECTMULTI(ALL).

5.1 ALLMULTI
The ALLMULTI keyword can cause multiple dumps of multivolume data sets to
be taken. Consider the situation where you have two jobs, one that dumps
VOL001 and VOL002, and the other that dumps VOL003. A large multivolume
data set is stored on VOL001 and VOL003. When both jobs include ALLMULTI,
the multivolume data set is included in the dump for volumes VOL001 and
VOL002 as well as in the dump for VOL003 (see Figure 9 on page 74).

Although this limited case can be ″handled″ by changing the volumes that the
jobs reference, in a real-life environment, this is often impossible to achieve. To
avoid missing any multivolume dumps, most installations using logical dumps
code ALLMULTI on each DFSMSdss job.

A more realistic example illustrates the difficulty in obtaining a single dump of a

data set when performing systemwide, logical dumps for disaster recovery
purposes:
• Eighty volumes are to be dumped as fast as possible to eight tape drives.
• Eight DFSMSdss jobs are to run.
• Each job is to reference 10 volumes.
• The installation makes extensive use of multivolume data sets, to the extent
that the minimum set of volumes required to ensure 100% coverage of all
the multivolume data sets is 80.

In the above situation, to ensure that each data set is dumped once, only one
DFSMSdss job can be run, which is impractical because:
• The total dump will be excessively long in terms of elapsed time.
• The total restore time will be unacceptably long in terms of restoring all of
the data within a disaster recovery window.

If multiple jobs are run:

 Copyright IBM Corp. 1997 73

 • The jobs will take longer than they need to because of the multiple dumps of
the same data set.
• There will be errors during the restore: The second and subsequent
restores of a data set will because of a ″duplicate data set″ condition.

DFSMSdss users want to select data sets during a ″logical volume dump″
according to the following criteria:
• The entire data set, if the first primary extent is found on any volume that is
referenced, even though not all volumes are on the list.
• The entire data set, if any part of the data set is found on any volume that is
referenced, even though all of the volumes are not on the list.

The first method of selection is what most DFSMSdss users really want and
need, but DFSMSdss until now has only supported the second method through
the ALLMULTI keyword.

5.2 SELECTMULTI
The SELECTMULTI keyword, is applicable during logical data set DUMP, data set
COPY, and conversion of volumes to or from SMS management. It is a
functional replacement for the ALLMULTI keyword. Because SELECTMULTI is
meaningful only when there is a volume list from which to select, SELECTMULTI,
like ALLMULTI, requires that either LOGINDDNAME or LOGINDYNAM be
specified. If SELECTMULTI is present but both LOGINDDNAME and
LOGINDYNAM are absent, existing message ADR138E is issued.

Figure 9. Problems with ALLMULTI. When two DFSMSdss jobs are run, one to dump VOL001 and VOL002, and
the other to dump VOL003, with both specifying ALLMULTI, the large multivolume data set is dumped twice.

74 DFSMS/MVS V1R4 Technical Guide

5.2.1 SELECTMULTI for DUMP and COPY
SELECTMULTI(ALL|ANY|FIRST) specifies the method for determining how
cataloged multivolume data sets are to be selected during a logical data set
dump or copy operation. SELECTMULTI is accepted only when logical volume
filtering is also used. Logical volume filtering is used when LOGINDDNAME or
LOGINDYNAM is specified during dump or copy. If logical volume filtering is not
used, the specification of SELECTMULTI is not accepted.
ALL is the default and indicates that DFSMSdss is not to dump or copy a
multivolume data set unless the volume list specified by
LOGINDDNAME or LOGINDYNAM lists all volumes that contain a part
of the data set or VSAM cluster. For VSAM data sets:
• If SPHERE is not specified, the data set in this context is any
cluster, base, or AIX.
• If SPHERE is specified, all parts of the base cluster and all AIXs
must be contained in the volume list.
ANY indicates that DFSMSdss is to dump or copy a multivolume data set
when any volume in the volume list specified by LOGINDDNAME or
LOGINDYNAM contains a part of the data set or VSAM cluster. For
VSAM data sets:
• If SPHERE is not specified, the data set in this context is any
cluster, base, or index.
• If SPHERE is specified, any part of the base cluster or any part of
any associated AIX may be contained in the volume list.
FIRST indicates that DFSMSdss is to dump or copy a multivolume data set
only when the volume list specified by LOGINDDNAME or
LOGINDYNAM lists the volume that contains the first part of the data
set. For VSAM data sets:
• If SPHERE is not specified, the volume list must include either the
volume containing the first extent of the data component for the
base cluster or the volume containing the first extent of the data
component of an AIX.
• If SPHERE is specified, the volume list must include the volume
containing the first extent of the data component for the base
cluster.

5.2.2 SELECTMULTI for CONVERTV

SELECTMULTI(ALL|ANY|FIRST) specifies the method for determining how
cataloged multivolume data sets are to be selected during conversion to or from
SMS management. The volume list is the list of volumes supplied by the
DDNAME or DYNAM keyword:
ALL is the default and indicates that DFSMSdss is not to process a
multivolume data set unless the volume list specified by DDNAME or
DYNAM lists all volumes that contain a part of the data set or VSAM
sphere.
ANY indicates that DFSMSdss is to process a multivolume data set when
any part of the data set or VSAM sphere is on a volume in the volume
list specified by DDNAME or DYNAM.

Chapter 5. DFSMSdss Multivolume Selection Enhancements 75

 FIRST indicates that DFSMSdss is to process a multivolume data set only
when the volume list specified by DDNAME or DYNAM lists the
volume that contains the first part of the data set or the primary data
component of the base cluster for a VSAM sphere.

5.2.3 Using the ADRUIXIT Exit

The setting of the SELECTMULTI keyword can be overridden by the DFSMSdss
installationwide Exit, ADRUIXIT, the Options Exit Routine, which is described in
DFSMS/MVS V1R3 Installation Exits .

5.2.4 Interactions
The specification of SELECTMULTI(ANY) has the same effect as the specification
of ALLMULTI on previous levels of DFSMSdss except when you convert a volume
to SMS. When converting a volume to SMS, specify SELECTMULTI(ANY) to
select a data set for processing if any part of the data set is on a volume in the
volume list.

The specification SELECTMULTI(ALL) has the same effect as the absence of the
specification of ALLMULTI on previous levels of DFSMSdss.

Specifying SELECTMULTI(FIRST) enables you, for logical data set DUMP, data set
COPY, and CONVERTV to non-SMS, to select a data set for processing only when
the first volume on which it resides is included in the volume list. The
specification of SELECTMULTI(FIRST) has the same effect as the specification of
ALLMULTI on previous levels of DFSMSdss when you convert a volume to SMS.

5.2.5 Errors
• If you specify ALLMULTI but not SELECTMULTI, SELECTMULTI(ANY) is set,
and a new message, ADR146I, is issued.
• If you specify both ALLMULTI and SELECTMULTI(xyz), SELECTMULTI(xyz) is
set, and message ADR146I is issued.
• If you alter the setting of ALLMULTI, either from OFF to ON or from ON to
OFF, the SELECTMULTI(xyz) set by the keyword is not altered, and a new
message, ADR147W, is issued.
• If the installation exit, ADRUIXIT, altered the setting of SELECTMULTI(xyz),
whether set by keyword or by default, the setting is altered and existing
message ADR035I is issued.

5.2.6 New Messages

The following messages are added in support of the SELECTMULTI changes:
ADR146I (xxx)-mmmmmm(yy), OBSOLETE KEYWORD keyword1 SPECIFIED.
keyword2 WILL BE USED
ADR147W (xxx)-mmmmmm(yy), INSTALLATION EXIT ATTEMPTED TO ALTER
OBSOLETE KEYWORD keyword. REQUEST IGNORED.
ADR148I (xxx)-mmmmmm(yy), MULTIVOLUME DATA SET dsname NOT
SELECTED.
ADR732E (xxx)-mmmmmm(yy), DATA SET dsname ON VOLUME
volume_serial_number WAS NOT SELECTED BECAUSE IT WAS NOT
CATALOGUED.

76 DFSMS/MVS V1R4 Technical Guide

 ADR878E (xxx)-mmmmmm(yy), THE FOLLOWING DATA SETS ON VOLUME
volume_serial_number {WILL NOT BE | WERE NOT} SUCCESSFULLY
PROCESSED dsname CATALOG: catalog_name
ADR886E (xxx)-mmmmmm(yy), DATA SETS EXIST ON VOLUME
volume_serial_number WHICH ARE NOT CONVERTIBLE {TO | FROM}
SMS MANAGEMENT, THE VOLUME {WILL BE | WAS} {PLACED |
LEFT} IN INITIAL STATUS

5.2.7 Migration and Coexistence

After you install DFSMS/MVS V1R4, you must change existing jobs to take
advantage of the multivolume selection enhancements in DFSMSdss. There are
coexistence considerations if the SELECTMULTI keyword is used in a job on a
system with a prior release of DFSMS/MVS installed.

5.2.7.1 Migration Considerations

To select a data set for processing when only the volume containing its first
extent is in the volume selection list, you have to change existing jobs to add
SELECTMULTI(FIRST).

Specification of ALLMULTI maps to the functionally equivalent

SELECTMULTI(ANY) unless CONVERTV to SMS is specified, in which case
ALLMULTI maps to SELECTMULTI(FIRST) so that you are not forced to change
existing jobs.

Absence of the specification of ALLMULTI maps to the functionally equivalent

SELECTMULTI(ALL) without forcing you to change existing jobs.

Warning message ADR147W is issued when the installation exit attempts to

modify the settings of ALLMULTI.

5.2.7.2 Coexistence Considerations

The SELECTMULTI keyword is not supported on releases previous to
DFSMS/MVS V1R4.

No other incompatibilities exist. Data sets dumped on releases that do not

support the SELECTMULTI keyword can be restored with DFSMSdss VR40. Use
of the new keyword during dump with DFSMSdss V1R4 does not introduce any
incompatibility during restore with a previous release.

No toleration PTFs are expected to be released for this announcement. The only
potential issue is in a shared spool environment with processors or LPARs
running different levels of DFSMS/MVS. Jobs executing DFSMSdss using the new
SELECTMULTI keyword must not be allowed to run on levels of DFSMS/MVS
previous to DFSMS/MVS V1R4.

5.2.8 Support Use Information

DFSMSdss manuals that describe or give examples of using the ALLMULTI
keyword have changed all descriptions and examples to use the SELECTMULTI
keyword together with the appropriate subparameter.

Chapter 5. DFSMSdss Multivolume Selection Enhancements 77

5.2.9 Performance
This enhancement to DFSMSdss does not change the performance of any
function. However, more efficient selection may result in a shorter elapsed time
for a group of data set dumps where the work is spread across multiple jobs. A
shorter elapsed time is possible because in many circumstances the same
multivolume data set could have been dumped multiple times with levels of code
before DFSMS/MVS V1R4.

5.2.10 Resources
No new control blocks have been added or deleted as a result of the
multivolume selection enhancement announcement. The sizes of existing
control blocks are not changed, and the announcement does not change the
DFSMSdss storage requirements.

78 DFSMS/MVS V1R4 Technical Guide

Chapter 6. DFSMSrmm

In this chapter we cover the following announcements and functional

enhancements for DFSMSrmm:
• Journal usage threshold
• Nonintrusive backup of the CDS
• Problem determination aid (PDA) trace
• Support for DFSMShsm alternate tape processing
• Recognition of external data managers
• Inventory management trial runs
Note: Some of the enhancements are not unique to DFSMS/MVS V1R4. They
are currently available through SPEs or APARs but are included in this chapter
for completeness.

6.1 Journal Usage Threshold

The DFSMSrmm journal data set contains a chronological record of changes to
the DFSMSrmm control data set. It allows the DFSMSrmm control data set to be
forward recovered. Before the journal usage threshold was available, the
journal could fill up unexpectedly.

As shown in Figure 10 on page 80, DFSMSrmm can now display information

messages to the operator, indicating how full the DFSMSrmm journal is
becoming. The first message to be displayed (EDG2107E) indicates that the
usage threshold, specified in the DFSMSrmm PARMLIB member, for this journal
has been reached. This message can be used to act as a trigger for a
procedure to be run automatically to back up both the control data set and
journal and then clear the journal. Subsequent messages are issued as the
journal becomes ever closer to becoming filled.

Customers have become concerned that they are not informed when the journal
is becoming full. When the journal is full, all tape processing must stop until the
journal and the control data set have been backed up and the journal cleared.
This procedure is disruptive and could involve a tape processing outage.

6.1.1 Update PARMLIB Member EDGRMMxx

Specify the JOURNALFULL threshold and BACKUPPROC parameters in
PARMLIB member EDGRMMxx on the OPTION command as follows:

6.1.1.1 JOURNALFULL(nn)
Specify JOURNALFULL to define a percentage-full threshold for the journal data
set. When DFSMSrmm detects that the journal has reached this threshold, it
issues message EDG2107E. DFSMSrmm also issues message EDG2107E at
DFSMSrmm startup if the journal has already reached the threshold specified. If
you specify a value of 0, DFSMSrmm issues no warnings on that system.
Different threshold values can be specified on systems sharing the RMM control
data sets.

 Copyright IBM Corp. 1997 79

Figure 10. Journal Usage Threshold. Messages are produced w h e n the threshold is reached and then every 5 %
until the 90% threshold is reached. Messages are then produced for every additional 1% usage.

Note: A backup procedure can be started automatically when the journal

percentage full threshold is reached. See BACKUPPROC for further information.
Specify a value in the 0-99 range. The default is 75.

6.1.1.2 BACKUPPROC(procname)
BACKUPPROC specifies the name of the procedure that you want started
automatically when the journal percentage full threshold is reached.

Specify a valid alphanumeric procedure name from one to eight characters.

There is no default value. If you do not specify a name, an automatic start
command is not issued.

To find out the setting of the threshold and the current journal utilization, either
issue this command:
RMM LISTCONTROL CNTL OPTION

or use the DFSMSrmm ISPF panels to show the System Options Panel Display.

If a journal is not allocated, the journal is disabled, or the utilization is less than
0.5%, a value of zero is returned for the current utilization value.

6.1.2 Automating Control Data Set Backup and Journal Clearing

You can use DFSMSrmm to automate the backup of the control data set and the
clearing of the journal. Although you can continue to run regular backups during
inventory management, automating control data set backup and clearing the
journal help ensure that the journal does not fill up. A full journal can impact
tape usage on your system.

To automate control data set backup and clearing of the journal you can:

80 DFSMS/MVS V1R4 Technical Guide

 • Specify a value for the JOURNALFULL operand on the OPTION command in
PARMLIB member EDGRMMxx as described in 6.1.1.1, “JOURNALFULL(nn)”
on page 79 or use the default value of 75%.
• Specify a procedure name for the BACKUPPROC operand as described in
6.1.1.2, “BACKUPPROC(procname)” on page 80. The submission of this
procedure is triggered when message EDG2107E is issued.
A procedure with the same name as specified on the BACKUPPROC parameter
must be stored in the system procedure library.

Figure 11 shows a sample procedure for backing up the control data set and
clearing the journal.

//EDGHSKP EXEC PGM=EDGHSKP,PARM=′ BACKUP(DSS)′

//MESSAGE DD DISP=SHR,DSN=RMM.MESSAGE
//SYSPRINT DD SYSOUT=*
//BACKUP DD DISP=(,CATLG),UNIT=TAPE,DSN=BACKUP.CDS(+1),
// LABEL=(,SL)
//JRNLBKUP DD DISP=(,CATLG),UNIT=TAPE,DSN=BACKUP.JRNL(+1),
// LABEL=(2,SL),VOL=REF=*.BACKUP
//DSSOPT DD *
CONCURRENT OPTIMIZE(1) VALIDATE
/*

Figure 11. CDS and Journal Backup and Clearing

Notes:
1. The names of the control data set and journal are obtained from the active
DFSMSrmm subsystem.
2. EDGHSKP backs up and clears the data set. If the R M M subsystem is not
active, you must use EDGBKUP, which does not clear the journal.
3. This example shows how DFSMSdss and DFSMSrmm work together to back
up the CDS and journal using concurrent copy; see 6.2, “Nonintrusive
Backup of the CDS” on page 82 for further information.

6.1.3 Messages
The following messages are displayed in relation to journal threshold
processing:
EDG2103E PERMANENT JOURNAL ERROR - REPLY ″R″ TO RETRY, ″ I″ TO
IGNORE, ″D″ TO DISABLE, OR ″L″ TO LOCK
• Look for a previous message with the EDG prefix that shows the
error.
• Notify your system programmer.
EDG2104E JOURNAL FILE IS FULL - SCHEDULE CONTROL DATA SET BACKUP
TO CLEAR IT
• Manually start the DFSMSrmm backup job to reset journal.
• Notify your system programmer.
• There is no reply for this message. This message is followed by
message EDG2103D, to which you must reply.

Chapter 6. D F S M S r m m 81
 EDG2107E JOURNAL THRESHOLD REACHED - JOURNAL IS percentage_value%
FULL. tracks TRACKS( kilobytesK) AVAILABLE
• The journal has reached the specified threshold value. If an
autostart procedure for backup is defined, RMM starts it
automatically. Otherwise follow your installation-defined backup
procedure.
• Notify your system programmer.
EDG2108E JOURNAL IS percentage_value% FULL. tracks TRACKS ( kilobytesK)
AVAILABLE
• This message is issued for every additional 5% full, or every 1%
once over 90% full. If a backup procedure has not been started,
follow your installation-defined backup procedure.
• Notify your system programmer.

6.1.4 TSO Subcommand Variables by Name

The variables in Figure 12 are available and can be used in REXX procedures
for managing the journal if required. For example a REXX procedure might be
developed that would display the current percentage usage of the journal.

┌───────────────────────────────────────────────────────────────┐
│Variable Subcommand Content Format │
│Name │
├───────────────────────────────────────────────────────────────┤
│EDG@JDS LC Journal name 44 characters│
├───────────────────────────────────────────────────────────────┤
│EDG@JRNF LC JOURNALFULL PARMLIB Numeric 0-99 │
│ operand value │
├───────────────────────────────────────────────────────────────┤
│EDG@JRNU LC Journal percentage Numeric 0-100│
│ used │
└───────────────────────────────────────────────────────────────┘

Figure 12. DFSMSrmm REXX Variables for the Journal

6.1.5 Migration and Coexistence

There are no migration or coexistence issues concerning the JOURNALFULL
operand and journal space utilization support.

6.2 Nonintrusive Backup of the CDS

The standard method of backing up DFSMSrmm′s control data set is to use the
DFSMSrmm backup utilities instead of other backup utilities, such as Access
Method Services EXPORT, because DFSMSrmm provides the necessary
serialization and forward recovery functions. DFSMSrmm backup utilities check
whether the control data set is in use, tell the DFSMSrmm subsystem that
backup or recovery is in process, and provide a way to forward recover using
the DFSMSrmm journal data sets.

Serialization prevents updates from occurring to the control data set and journal
until the backups are complete, thus ensuring the integrity of the backups so that
recovery is possible. While the backups are taking place, however, any tape

82 DFSMS/MVS V1R4 Technical Guide

 processing involving updates to the DFSMSrmm CDS or journal has to wait until
the backups are complete. If the control data set is large, this outage could last
for an unacceptably long time.

DFSMSrmm now allows an installation to use concurrent copy to minimize the

outage. With using concurrent copy, the backups are logically complete in just a
few seconds, after which normal DFSMSrmm tape management functions can
continue while the physical dumps take place. Concurrent copy enables
nonintrusive backup of the DFSMSrmm control data set as follows:
• The control data set and journal can be backed up without impact on tape
use.
• Updates resulting from tape use or from commands are permitted when
backup is in progress and do not affect the integrity of the resulting backup
copy.
• Backup may now be made directly to tape.
Note: This is true even if concurrent copy is not being used.

6.2.1 Functional Characteristics

The backup utilities of DFSMSrmm have been enhanced such that DFSMSdss
can be invoked instead of Access Method Services. This is achieved by coding
BACKUP(DSS) in the parm field of the JCL′s EXEC statement, and by optionally
coding a DSSOPT DD statement. The DSSOPT statement allows the installation
to code any appropriate parameters that would normally be coded on the
DFSMSdss DUMP and RESTORE control cards. The DSSOPT DD statement can
be used to specify customized operands for the DUMP and RESTORE commands
used by DFSMSrmm. You might customize the operands according to the media
used for the backup and the resources that are available. The operands
specified with the DSSOPT DD statement override the default DFSMSdss DUMP
and RESTORE commands issued by DFSMSrmm. DFSMSrmm reads all the
records specified with the DSSOPT DD statement and uses them to replace its
default command operands beginning at the second line.

The default DFSMSdss DUMP and RESTORE commands issued by DFSMSrmm

when the EDGBKUP EXEC parameter BACKUP(DSS) is specified are:
DUMP DS(INCLUDE(cds_name)) OUTDD(BACKUP) SHARE -
COMPRESS CONCURRENT VALIDATE OPTIMIZE(1)

RESTORE DS(INCLUDE(**)) INDD(BACKUP) -

REPLACE

Figure 13 on page 84 shows an example of using the DSSOPT DD statement.

Chapter 6. D F S M S r m m 83
 //EDGHSKP EXEC PGM=EDGHSKP,PARM=′ BACKUP(DSS)′
//MESSAGE DD DISP=SHR,DSN=RMM.MESSAGE
//SYSPRINT DD SYSOUT=*
//BACKUP DD DISP=(,CATLG),UNIT=TAPE,DSN=BACKUP.CDS(+1),
// LABEL=(,SL)
//JRNLBKUP DD DISP=(,CATLG),UNIT=TAPE,DSN=BACKUP.JRNL(+1),
// LABEL=(2,SL),VOL=REF=*.BACKUP
//DSSOPT DD *
CONCURRENT OPTIMIZE(4) VALIDATE COMPRESS
/*

Figure 13. DSSOPT DD Statement

The DUMP command operands you can specify in the DSSOPT DD statement are
controlled and validated by DFSMSdss, not by DFSMSrmm. If unsupported
command operands are specified, DFSMSdss fails the dump operation.
Comments can be included in the DSSOPT records, but they must be comments
acceptable to DFSMSdss. Refer to the DFSMS/MVS V1R3 DFSMSdss Storage
Administration Reference for command operands that are supported.

The DFSMSdss parameter keywords specified in the DFSMSrmm housekeeping

job in Figure 13 have the following meanings:
CONCURRENT specifies that the data is to be processed with concurrent copy if
possible; otherwise, the data will be processed as if CONCURRENT
were not specified.
OPTIMIZE(4) specifies that the data is to be dumped as fast as possible by
reading a cylinder′s worth of data at a time and overlapping the
reading of the data with writing the previous cylinder′s worth of
data to tape.
VALIDATE specifies that all indexed VSAM data sets are to be validated as
they are dumped.
COMPRESS specifies that the dumped data is to be written in compressed form
to the output medium. In this example the output medium is tape.
Even though the output tape may support hardware compression
performance will benefit from using COMPRESS as less data will be
transferred across the channel.

6.2.2 Migration and Coexistence

When DFSMSdss is used as the backup method, DFSMSrmm must run with
journaling active. Backups taken using the new BACKUP(DSS) option cannot be
restored using previous levels of DFSMSrmm code. However, a backup taken by
DFSMSrmm using DFSMSdss can be restored directly with the ADRDSSU utility
and the EDGBKUP utility for forward recovery from the journal and journal
backups.

Backups taken on either level of code that do not use DFSMSdss are eligible on
either level of code.

When BACKUP(DSS) is requested, updates to the control data set are allowed
during journal backup. In addition, updates during control data set backup are
allowed if concurrent copy is used. The journal backup includes records that are
not included in the control data set backup. For this reason the restore process
now involves forward recovery from the latest journal backup as well as the

84 DFSMS/MVS V1R4 Technical Guide

 active journal. If only the active journal is used, some records may not be
recovered. EDGBKUP detects this situation and issues message EDG6424E.

To use DFSMSrmm nonintrusive backup, an installation must have a DASD

subsystem controller that supports concurrent copy. The IBM 3990-6 DASD
subsystem controller supports concurrent copy, as do the IBM 3990-X03
controllers with the Extended Platform.

Using DFSMSdss with concurrent copy can dramatically reduce the time during
which DFSMSrmm is unavailable for tape management processing. The dump
output is fully compatible with dumps produced without concurrent copy.

Dumps produced on a system using concurrent copy can be restored on systems

that do not support concurrent copy.

6.2.3 Performance
Although DFSMSrmm nonintrusive backup does not speed up the actual dump, it
dramatically reduces the time during which DFSMSrmm is unavailable for
performing tape management functions. With concurrent copy, DFSMSrmm is
unavailable for only a few seconds rather than the minutes before the option was
available.

6.3 Problem Determination Aid Trace

The DFSMSrmm PDA trace provided with DFSMS/MVS V1R4 will gather
diagnostic data at specific internal module points and record it in a circular file
within storage and (optionally) on DASD (see Figure 14 on page 86). The
purpose of the trace is to gather sufficient information during DFSMSrmm
execution to enable subsequent analysis to pinpoint module flow and resource
usage related to any given problem.

From a customer perspective, PDA trace support addresses the serviceability

issue of first-time data capture. Currently, if a DFSMSrmm defect occurs, it
requires multiple re-creations, traps, and dumps to understand the failing
scenario. The PDA trace decreases the amount of time to diagnose defects
should they occur.

6.3.1 Functional Characteristics

The overall design objective of the PDA trace is to maximize data capture with
minimum performance impact. The PDA trace facility consists of the following:
• Entry and exit trace requests throughout DFSMSrmm
• A trace data accumulation function
• A trace data recording function

The DFSMSrmm trace recording function receives the trace data scheduled for
output and writes it to a file on DASD. The PDA trace consists of two separate
log data sets. DFSMSrmm recognizes these log data sets by their DD names,
EDGPDOX and EDGPDOY. Recording takes place in the data set defined by
EDGPDOX. When that data set is filled, the two data set names are swapped,
and recording continues on the newly defined data set.

When the newly defined data set is filled, the names are again swapped, and the
output switches to the other data set, thus overlaying the previously recorded

Chapter 6. D F S M S r m m 85
Figure 14. DFSMSrmm Problem Determination Aid

data. The larger the file, the longer the period of time that will be represented
by the accumulated data.

The preferred implementation of the PDA trace is to establish a protocol that

automatically copies the EDGPDOY data set to tape as a GDG data set each time
message EDG9117I is issued. This practice provides a sequential history of trace
data over time so that the data is available when needed for resolving problems.
Sample JCL will be provided for dumping this trace data as a series of GDGs.

The in-storage circular file is a contiguous area of storage divided into blocks.
Each block contains multiple variable-length trace entries. Each block is written
to DASD as a separate unit of data. The size of each block and number of
blocks in the file are controlled by user-defined values within an in-storage
circular file. The DASD block size for the EDGPDOX/EDGPDOY data sets and the
number of blocks or buffers that make up the trace wrap table can be changed
with PARMLIB options:
PDABLKSZ(min=1, max=31) and
PDABLKCT(min=3, max=255)

The total size of the in-storage trace wrap table is the product of these two items
in kilobytes (KB). For example, if you specify PDABLKSZ(4) and PDABLKCT(12),
you get a total in-storage trace table of 48KB. The default in-storage trace wrap
table sizes if PDABLKSZ and PDABLKCT are not specified are based on where
the RMM trace data sets are placed. These defaults are:
EDGPDOX on 3380 = 5610K
EDGPDOX on 3390 = 6885K
default if no EDGPDOX = 3060K

The PDA trace came from DFSMShsm. The reuse of this trace facility will allow
consistency in how the two products implement a trace facility. Customers and
support groups familiar with the DFSMShsm trace facility will be able to use the

86 DFSMS/MVS V1R4 Technical Guide

 DFSMSrmm trace facility without having to learn anything radically new. Use of
the DFSMSrmm trace facility requires no new keywords.

6.3.1.1 Invocation
The trace facility is enabled at the end of the DFSMSrmm startup unless
specifically inhibited by command in the startup procedure.

The default operating mode for the trace facility is:

• PDA ON
• PDALOG ON
• PDALOG SWAP PDOX and PDOY at initial startup
• PDABLKCT(255)
• PDABLKSZ(22) for 3380, 27 for 3390, or 12 for any other device

Typical commands that can be used from the console to control the DFSMSrmm
PDA trace are:
F DFRMM,PDA=ON (turns on tracing)
F DFRMM,PDALOG=SWAP (used to SWAP the DASD trace data sets)

Frequently, only several minutes to one hour of the PDA trace will be required to
analyze a problem. To reduce the amount of data to be sent for analysis, use
the DFSMSrmm trace formatter program, ARCPRPDO, to select and copy all
trace entries created during the time of interest.

6.3.2 Migration and Coexistence

If an installation does not update its DFSMSrmm startup procedure to include the
new DD EDGPDOX and EDGPDOY statements, it will receive informational
message EDG9115I, and the tracing will be recorded only in the in-storage trace
buffer.

6.3.3 Support Use Information

Two new data sets are required for PDA trace output data logging to DASD.
They are not required if trace data is not to be written to DASD. These data sets
are physical sequential, in variable-length record format, with a maximum record
length of 256 bytes and a maximum block size of 31748 bytes. The actual block
size is controlled by PARMLIB options or the default values. The data sets
should have a disposition of SHR so that they can be examined online,
concurrent with DFSMSrmm operation. Because of the data set swap and
rename process, both the EDGPDOX and EDGPDOY data sets must reside on the
same volume. Care must therefore be taken if these data sets are to be
allocated to SMS managed volumes. In a multihost environment, each different
host must have its own unique EDGPDOX and EDGPDOY data sets. The PDA
output data sets must not be shared with any other system component for trace
data accumulation.

In the DFSMSrmm startup procedure, the following new DD statements are

required for the EDGPDOX and EDGPDOY data sets:
//EDGPDOX DD DSN=hlq.RMMPDOX,DISP=SHR
//EDGPDOY DD DSN=hlq.RMMPDOY,DISP=SHR

Chapter 6. D F S M S r m m 87
 IEFBR14 could of course be used to preallocate these data sets. An initial size
recommendation for these data sets is 20 cylinders. Then adjust them according
to installation requirements.

When DFSMSrmm switches the output data sets (indicated by message

EDG9117I), the trace data before the switch is available in the data set defined by
the EDGPDOY DD statement.

The DFSMShsm PDA formatter (ARCPRPDO utility) can be used to format the
EDGPDOX and EDGPDOY files. Users are not required to have the DFSMShsm
license to use the ARCPRPDO utility to format the DFSMSrmm PDA files.

6.3.4 Performance
Although the trace circular file processing is performed inline, the logging to
DASD is an independent task. This task is not merged inline with DFSMSrmm
function, has minimal dependence on the product structure or design, and
competes separately for time and resources. Therefore, the trace′s potential for
degrading DFSMSrmm performance and reliability is minimal.

6.4 Support for DFSMShsm Alternate Tape Processing

If APAR OW21287 is not installed, DFSMSrmm prevents an installation from using
the alternate copy of a backup or migration tape to recall or recover a user data
set. DFSMSrmm checks that the full 44-character DFSMShsm data set name
matches the name recorded in DFSMSrmm′s control data set. Because the data
set name of the alternate tape has COPY as its second qualifier, it fails the
check.

To work around this, an operator issues the RMM CHANGEVOLUME

subcommand to change the data set name known to DFSMSrmm to the alternate
data set name. DFSMSrmm then allows the alternate tape to be used by
DFSMShsm.

The commands are:

RMM CHANGEVOLUME altvol INIT(Y) CRLSE(INIT)
RMM CHANGEVOLUME altvol DSN(′ prefix.xxxxTAPE.DATASET′ )

where:

altvol
specifies the volser of the copy that is being used to replace the original

prefix
specifies the DFSMShsm-defined backup or migration prefix

xxxx
specifies either BACK or HMIG, depending on the DFSMShsm tape type

When APAR OW2157 is installed, DFSMSrmm validates only the last 17

characters of the data set name used by DFSMShsm when it reads the tape for
input processing. This DFSMSrmm validation allows DFSMShsm to use the
alternate tape without having to inform DFSMSrmm about the change of data set
name to be used.

88 DFSMS/MVS V1R4 Technical Guide

6.5 Recognition of External Data Managers
DFSMSrmm can now be instructed to manage a tape volume but not to track all
the data sets that are written to that volume. The EDGUX100 installation exit
now supports a new function: RMM records data set details for only the first file
on a tape volume but keeps track of the statistics at the volume level.

Customers may have applications that manage their own pool of tapes and the
data sets on these tapes. Until now these applications (known as external data
managers to some other tape management system) have stored information
about these data sets, and DFSMSrmm has duplicated it in its control data set.
Thus more disk space was required (for the data set information being
duplicated), and the backup for DFSMSrmm′s control data set and journal was
taking longer than necessary.

When DFSMSrmm is told through the EDGUX100 installation exit to record only
information for the first file on a tape volume, it still performs normal volume
validation, but it can no longer perform data set name checking for the second
and subsequent files on the volume.

DFSMSrmm can still be used to manage the volume based on the first file details
and the statistics maintained at the volume level.

6.5.1.1 Requesting Recording of First File Data Set Details

This option is available to the EDGUX100 installation exit only during the OPEN
processing for the first file on a volume. The sample exit retrieves the data set
name, job name, and job step program name from the system control blocks,
before scanning a table for a match. If there is a match, the sample exit sets the
PL100_SET_IGNORE_FILE2_TON bit to request that DFSMSrmm maintain data set
details only for the first file.

6.5.1.2 Updating the EDGUX100 Exit

To use the EDGUX100 exit for this function, update the sample table with the job
name, data set name, and program name for each application that can manage
its own data set inventory on tape. To make the job name, data set name, and
the program name generic, include either a % or an *. The % represents a
positional character, and the * can be used to tell the exit to ignore all remaining
characters in any of the three pieces of information per table entry.

Figure 15 on page 90 shows a sample EDGUX100 installation exit with the table
coded to include the names of four external data managers.

Chapter 6. D F S M S r m m 89
 EDMTAB DS 0F *START OF TABLE
* JOBNAME DATA SET NAME
*
DC CL8′ * ′ , CL44′ BACKUP*′
DC CL8′ ABC*′ PROGRAM NAME
*
DC CL8′ STSGWD* ′ , CL44′ *′
DC CL8′ A*′ PROGRAM NAME
*
DC CL8′ STSG%D* ′ , CL44′ STSG%%.BACKUP.*′
DC CL8′ DEF%MAIN′ PROGRAM NAME
*
DC CL8′ STSGDPW ′ , CL44′ DAVE.TOOMUCH.DATA.*′
DC CL8′ AB999*′ PROGRAM NAME
*
DC CL8′ EDMEND′ *END OF TABLE MARKER

Figure 15. EDGUX100 Installation Exit. The table entries are coded to include four external data managers.

Using the EDGUX100 exit to specify external data managers allows quite a lot of
flexibility in specifying the type of jobs, data sets, and programs for recording
only the first file on a tape volume. For example, the first entry in Figure 15 asks
the exit to look for all data sets with a high-level qualifier starting with the
characters BACKUP that is being written by any program starting with the
characters ABC. In this particular example, the job name is specified with an *,
so all jobs would be considered.

6.6 DFSMSrmm Inventory Management Trial Runs

It is now possible to check what DFSMSrmm will do during VRSEL processing in
a trial run of inventory management vital record processing, without DFSMSrmm
actually making any changes to the control data set or journal. If required, this
trial run can be made for any specified date in the future. The trial run lets the
installation see how the vital record specifications, which have been defined or
changed, would be processed in a production run, except that DFSMSrmm does
not make any changes to the control data set. DFSMSrmm can be set up to
require that tape volumes and data sets are ″covered″ by a specified minimum
number of vital record specification (VRS) policies. If an installation has added,
changed, or deleted VRS policies, the system can be set up such that a
production inventory management run cannot be performed until a successful
trial run has been made.

To reduce the possible exposures occurring from changes made to VRS policies,
such as causing changes to tape data set and volume expiration processing,
customers have requested the following changes to DFSMSrmm:
• A trial run capability for inventory management
• A report of updates made by inventory management
• The option to enforce a trial run if policy changes are made
• The provision of thresholds for inventory key actions

This release of DFSMSrmm addressed each of the above requirements although

threshold processing has been restricted in this release to checking that a
minimum number of VRS policies cover the tape data sets and volumes. This
will catch the situation where either several VRS policies have been deleted by
mistake or an error on the control data set has prevented all policies from being

90 DFSMS/MVS V1R4 Technical Guide

 read into storage. In the past such mistakes or problems could have led to a
large number of data sets and/or tape volumes expiring as they would no longer
be covered by a valid VRS.

Now that the date can be specified on the trial inventory management run, an
installation can check what effect the existing or changed VRS policies would
have at some time in the future. For example, it would be possible to show how
many tapes would be moved offsite after month end processing had completed.
It would also be possible to see whether sufficient scratch tapes would be
released and available for use on the weekend.

6.6.1 Inventory Management Processing

If VRSCHANGE(VERIFY) is specified in the EDGRMMxx PARMLIB member and
changes have been made to retention and movement policies since the last
successful run of inventory management, any attempt at running VRSEL (without
VERIFY also being specified), DSTORE, or EXPROC processing will fail. If VRSEL,
DSTORE, or EXPROC is requested and changes have been made to retention
and movement policies since the last successful run of inventory management,
message EDG2308I is issued to the MESSAGE file regardless of the VRSCHANGE
option. If a trial inventory management run is requested (VERIFY is coded as a
parameter on the EXEC statement), neither DSTORE nor EXPROC can be
specified; otherwise processing stops and message EDG6002E is issued.

The ACTIVITY file is now supported, and, during VRSEL processing, details of
changes made to data set information are added to it. If the ACTIVITY file is not
allocated for a VERIFY run, existing message EDG6101E is issued, and
processing fails.

Figure 16 on page 92 shows what you must do to set up the environment for
inventory management trial runs.

Chapter 6. D F S M S r m m 91
Figure 16. Inventory Management Trial Runs

Once a successful inventory management trial run has completed (VERIFY was
specified as a parameter on the EXEC statement), a production inventory
management job can be submitted, even if the VRSCHANGE PARMLIB option
specifies VERIFY, because DFSMSrmm knows that no changes to VRS have been
made.

6.6.2 Inventory Management VRSEL Processing

During processing DFSMSrmm makes no control data set changes during a
VERIFY run, other than to show that VERIFY has been run successfully.

DFSMSrmm now counts all VRSs (volume, data set name, and name) that it does
not delete and writes this number with message EDG2229I to the MESSAGE data
set. DFSMSrmm also compares this number to the number specified through the
VRSMIN PARMLIB option and uses the VRSMIN action value (FAIL, WARN, or
INFO) to decide whether, and how, processing is to continue.

If the ACTIVITY file is allocated and open, details of changes to data set records
and their vital record status are added to the file during processing.

When a NEXTVRS is missing (VRS chain is broken), a new message, EDG2230I,

is issued. If VRSCHANGE(VERIFY) is specified in the EDGRMMxx PARMLIB
member, this NEXTVRS chain must be repaired and a new VERIFY job run
successfully before a production inventory management job can be run.

92 DFSMS/MVS V1R4 Technical Guide

6.7 Migration and Coexistence
When converting to DFSMS/MVS V1R4, consider the following points:
• Several new messages are issued to the MESSAGE file during inventory
management (see 6.7.3, “New and Changed Messages” on page 94 for
details).
• The default for the VRSCHANGE PARMLIB option is that VERIFY is required.
Therefore, unless this default is changed to INFO, an inventory management
trial run will be required following any changes made to VRS, before the next
production run for inventory management is made.
• The default for VRSMIN is that one VRS is required. The recommendation is
that VRSMIN be coded as a PARMLIB option to a more realistic value on the
basis of how many VRS policies exist for the installation. If changes to VRS
policies are made such that they expire or become deleted, this number
should be set to a number lower than the current number being used and
periodically reviewed.
If a VRSMIN value is defined and that number of VRSs does not exist,
processing stops, unless an alternative action is specified (see 6.7.2, “New
PARMLIB Options” for details of coding this operand).

6.7.1 New EXEC Parameters for EDGHSKP parameters

DFSMSrmm provides the EDGHSKP utility to help you perform inventory
management. The following new EDGHSKP EXEC parameters are provided in
support of inventory management trial runs:
DATE(date|+nnn) The date to be used for VERIFY processing. Dates can be
provided in the format yyddd, such as 98030, or yyyy/ddd, such
as 1998/030. Both formats indicate 30 January 1998. For the
year 2000 and higher, the yyyy/ddd format must be used.
The alternative is to enter the number of days to add to the
current date to determine the actual date to be used for VERIFY
processing. For example, DATE(+7) tells DFSMSrmm to use
the date of seven days in the future.
The default is the current date.
VERIFY Specify this parameter to request that DFSMSrmm not update
the control data set on this run of inventory management. This
parameter is supported only for VRSEL processing. Neither
DSTORE nor EXPROC can be specified along with VERIFY.
Note: With VRSCHANGE(VERIFY) specified in the PARMLIB
option statement, only a single, successful VERIFY run is
required to enable full inventory management processing. If
the expected results are not achieved despite a successful run,
the policies must be modified and further VERIFY runs made
until results are acceptable.

6.7.2 New PARMLIB Options

The following options are added to the PARMLIB member EDGRMMxx in support
of inventory management trial runs:
VRSMIN(count,action) Use this operand to specify a minimum number of VRSs
and the action to be taken by DFSMSrmm when the
minimum number of VRSs is not defined.

Chapter 6. D F S M S r m m 93
 Also, to control what DFSMSrmm should do when the
count is not reached, specify one of the following:
FAIL Issue message EDG2229I to the MESSAGE file
and stop inventory management processing.
A return code of 8 is set. This is the default.
WARN Issue message EDG2229I to the MESSAGE file
and continue processing. A return code of 4 is
set.
INFO Issue message EDG2229I to the MESSAGE file
and continue processing.
VRSCHANGE This operand determines which action DFSMSrmm should
take during inventory management following changes to
VRS policies using ADD or DELETE subcommands.
Optionally one of the following can be specified:
INFO Changes to VRS policies will not force a
VERIFY run to be made
VERIFY VRS policy changes must be verified by
running EDGHSKP vital records selection with
the VERIFY parameter. This program must run
successfully before a production run can be
performed.
VERIFY is the default

6.7.3 New and Changed Messages

The following new and changed messages have been introduced in support of
inventory management trial runs:
EDG6101E REQUIRED DDNAME ddname NOT SPECIFIED
EDG2229I NUMBER OF VRS RECORDS READ IS number
EDG2230I NEXTVRS name_vrs DOES NOT EXIST. CHAINING vrs_type VRS IS
vrs_mask.
EDG2308I CHANGES HAVE BEEN MADE TO VRS POLICIES SINCE THE
PREVIOUS INVENTORY MANAGEMENT RUN
EDG2309I THE PARMLIB OPTIONS CURRENTLY IN USE ARE
VRSJOBNAME( job_option) VRSMIN( count_value,action_value )
VRSCHANGE( change_option ) CATRETPD(hours)
EDG2310I INVENTORY MANAGEMENT STOPPING BECAUSE OF
VRSMIN( count_value,FAIL) OPTION
EDG2311I INVENTORY MANAGEMENT STOPPING BECAUSE OF
VRSCHANGE( change_option ) OPTION
EDG2312I CHANGES HAVE BEEN MADE TO VRS POLICIES SINCE THE START
OF THE VERIFY RUN - A FURTHER VERIFY RUN IS REQUIRED
EDG6001I INVENTORY MANAGEMENT STARTING ON date AT time -
PARAMETERS IN USE ARE parameters

94 DFSMS/MVS V1R4 Technical Guide

Chapter 7. Distributed Data Access

In today′s distributed computing environment, applications must often access

data residing on different computers in a network. In many cases, the most
effective data access services occur when applications can access remote data
as if it were local data.

To provide such services, DFSMS/MVS offers the following solutions:

• Network File System Server, which supports client platforms using TCP/IP
communication protocols and NFS file access protocols
• Distributed FileManager/MVS, which supports client platforms using APPC
communication protocols and DDM file access protocols

These data access services enable users and applications on heterogeneous

client computers in your network to take advantage of storage resources on
MVS/ESA and OS/390 systems, including system-managed storage,
high-performance storage access, file access security, data sharing, and
centralized data access.

7.1 Network File System

The DFSMS/MVS Network File System is an MVS/ESA implementation of the
industry standard Network File System (NFS) Version 2 protocol originally
created by Sun Microsystems, Inc.. The DFSMS/MVS NFS server enables your
MVS/ESA system to act as a file server or client to any authorized NFS clients on
a TCP/IP network.

The NFS protocols have become one of the most pervasive methods for
interoperability between different types of operating systems. The NFS protocol
is designed to allow transparent access between a client system and the server
system, with the files contained on the server system appearing as if they were
available locally on the client system. The NFS client program on the client
system maps local file system calls into network calls and sends a request for
data or control information to the NFS server. In this way, NFS client and server
deliver remote data access to an application without special interfaces in the
application.

The files from the server are mounted on the client′s file system. Data transfer
occurs when files are read or written and the data is transferred using TCP/IP
facilities across the network. A client can see much more data than can be held
locally, and many clients can share the same data, thus avoiding distribution of
data, downloading, and needless duplication.

7.2 DFSMS/MVS Network File System

IBM′s DFSMS/MVS NFS feature adds significant capabilities that enable OS/390
and MVS systems on a TCP/IP network to access data on other NFS server
machines (for example, OS/390, RS/6000, SUN, HP, and PCs using OS/2).

Several functional changes were introduced with DFSMS/MVS V1R3 NFS, which
was announced in September 1996:
• Provide DFSMS/MVS NFS client support

 Copyright IBM Corp. 1997 95

 • Retrieve the attributes of a migrated data set without recall

7.2.1 DFSMS/MVS NFS Client Function

Evolving from a UNIX environment, many systems support both NFS server
capabilities and NFS client capabilities. For some time, the DFSMS/MVS product
has had NFS server support as a feature. NFS client support will now be
included in the DFSMS/MVS V1R3 NFS feature for the OS/390 and MVS
platforms. The DFSMS/MVS NFS client program supports the SUN NFS Version 2
protocols. It uses remote procedure call (RPC) requests to communicate with
remote NFS servers over a TCP/IP network (see Figure 17). Thus one OS/390
(or MVS) system can exchange data files with another using the NFS
client/server technology. The DFSMS/MVS NFS client provides the full capability
for OpenEdition applications to read or write MVS conventional data sets and
OpenEdition Hierarchical File System (HFS) files on another OS/390 system that
has the DFSMS/MVS NFS feature installed.

With the DFSMS/MVS NFS client support, DFSMS/MVS has the complete client
and server implementation of NFS for MVS.

Figure 17. DFSMS/MVS NFS client feature. The NFS MVS client allows you to read and write data on UNIX and
other MVS machines.

Some of the functions supported by the DFSMS/MVS NFS client include:

• Create and delete files
• Read and write files
• List directories
• NFSSTAT and SHOWMOUNT commands
• Support for UNIX-style credentials authentication, such as UID, GID, and
permission bits

96 DFSMS/MVS V1R4 Technical Guide

 • Additional Security Authorization Facility (SAF) protection to a remote
DFSMS/MVS NFS server with the MVSLOGIN command
• Support for simple data conversion of single-byte character sets

7.2.1.1 Configuration Tasks

The tasks to configure the DFSMS/MVS NFS client are listed below. For detailed
parameter descriptions refer to the manual DFSMS/MVS V1R3 NFS
Customization and Operation , SC26-7029.
1. Note DFSMS/MVS NFS client considerations
The DFSMS/MVS NFS client does not share the same address space with
OpenEdition MVS. During OpenEdition MVS file system initialization, the
DFSMS/MVS NFS client is started and run in the Logical File System (LFS)
colony address space.
2. Set up the DFSMS/MVS client authorization
All programs that come with the DFSMS/MVS NFS client must reside in an
authorized program facility (APF)-authorized program library. The IEAAPFxx
PARMLIB member must be updated with the DFSMS/MVS NFS client library
name defined in the GFSCPROC STEPLIB. A sample GFSCPROC is provided
as a member of library NFSSAMP.
3. Define the DFSMS/MVS NFS client to RACF
Define a RACF userid for the DFSMS/MVS NFS client. Because the client
runs as a started task, you must define an entry in the RACF-started
procedures table that associates the DFSMS/MVS NFS client userid with the
client startup procedure name.
4. Update the IFAPRDxx or IGDDFPKG PARMLIB member
For OS/390 Enable/Disable support add the following to the IFAPRDxx
member of PARMLIB:
PRODUCT OWNER(′ IBM CORP′ )
NAME(′ OS/390′ )
FEATURENAME(′ DFSMS/MVS NFS′ )
VERSION(*)
RELEASE(*)
MOD(*)
ID(′5645-001′ )
STATE(ENABLE)
If you are installing DFSMS/MVS V1R3 or higher on a system other than
OS/390, add the following statement to the IGDDFPKG member of PARMLIB
to enable the DFSMS/MVS NFS feature product:
DFSMS_FEATURE=(systemid.NFS)
5. Update the OpenEdition MVS PARMLIB member
The NFS Client is started during OpenEdition MVS initialization. The
FILESYSTYPE statement must be present in OpenEdition PARMLIB member
BPXPRMxx . It defines the type of file system.
FILESYSTYPE
TYPE(NFS) <----- NFS Client
ENTRYPOINT(GFSCINIT) <----- Entry point for NFS Client
initialization
PARM(′ installation parms′) <----- Installation parameters for
NFS Client

Chapter 7. Distributed Data Access 97

 ASNAME(proc_name) <----- Procedure name in
SYS1.PROCLIB to start
colony address space
The name of the TYPE operand must be NFS; otherwise the utility program
nfsstat may fail.
The ASNAME(proc_name) operand specifies the procedure name in
SYS1.PROCLIB that is used by OpenEdition MVS to start the address space
in which the DFSMS/MVS NFS client will be initialized. The proc_name is
also used for the name of the address space.
6. Update SYS1.PROCLIB
Add the procedure name specified in the ASNAME(proc_name) operand to
the system PROCLIB.
The DFSMS/MVS NFS client is required to start in a separate and
stand-alone LFS colony address space for data integrity and data isolation
among different physical file systems (PFSs). The sample DFSMS/MVS NFS
client startup procedure can be found as member GFSCPROC in the
NFSSMP library.
7. Allocate DFSMS/MVS NFS client log data sets
The NFS stores messages in the DFSMS/MVS NFS client data sets specified
in the NFSCMSG1 and NFSCMSG2 statements of the clients startup
procedure. The log data sets must be preallocated with the following
attributes:
• DSORG=PS
• RECFM=VB
• LRECL=137
• BLKSIZE=6144
8. Mount remote file systems
OpenEdition does not support NFS mounts in the SYS1.PARMLIB member
statement. You can use the OpenEdition MVS automount facility or the TSO
MOUNT command to make a connection between a mount point on your
local MVS HFS and one or more files on a remote MVS, AIX, UNIX, OS/2,
OS/390, or other file system.
The remote file system must be mounted on the OpenEdition MVS file
system before any reference is made to the remote data. Once mounted, the
remote file system can be treated as an extension of the local OpenEdition
MVS file system.

7.2.2 OpenEdition Interface

The DFSMS/MVS NFS client support operates under OpenEdition (UNIX
Services). Thus applications can use the OpenEdition POSIX-compliant system
interfaces to utilize the NFS client functions with few changes to those
applications. In addition conventional DFSMS/MVS Basic Sequential Access
Method (BSAM), Queued Sequential Access Method (QSAM), Virtual Storage
Access Method (VSAM) ESDS applications, as well as OpenEdition applications
and users, have transparent access to data on any remote NFS server platform
supporting SUN NFS Version 2 protocols.

The NFS Client is implemented on OpenEdition MVS as a new PFS. It

implements the client portion of the NFS Version 2 protocols and uses TCP/IP to
communicate with a remote NFS server.

98 DFSMS/MVS V1R4 Technical Guide

 7.2.2.1 Access Method Support for OpenEdition Files
DFSMS/MVS V1R3 access methods offer a new opportunity to access enterprise
data in an open systems environment. Many applications written to use the
BSAM/QSAM access methods, and some applications written to use the VSAM
ESDS access, now have transparent access to files associated with OpenEdition.
With this support, the MVS conventional application using the normal record or
block interfaces provided by the DFSMS/MVS access methods can access files
residing in remote NFS server platforms. This support will map between the
record interface or block interface and the byte-oriented structure of the
OpenEdition or remote files. The following files can be accessed by specifying
the PATH=pathname parameter on the BSAM, QSAM, or VSAM JCL DD
statement or SVC 99:
• Local OpenEdition files
OpenEdition regular files are managed by the DFSMSdfp HFS. Each HFS file
is an identified unit of text or binary data and is byte-oriented (versus
record-oriented). The HFS supports file names and pathnames up to 255
characters long. Within the HFS file name, special characters and
lower-case characters can be embedded.
• Remote files
A remote file is any regular file residing in a remote NFS server platform that
supports the SUN NFS Version 2 protocols. With this function, MVS
applications written to use BSAM, QSAM, and VSAM ESDSs can access
remote files through the DFSMS/MVS NFS client and the network
environment just as if the remote files were local.
• Local FIFO special files
An FIFO (First-In-First-Out) special file, or named pipe, is an OpenEdition
special file. An FIFO special file is defined in a DD statement with both
PATH= and DSNTYPE=PIPE. An FIFO special file provides data
connections between programs and typically sends data from one program
(or process) to another. Both ends of an FIFO special file can be used in a
single program task. All data in an FIFO special file vanishes when the last
program using it closes it.
• Local character-special files
A character-special file is an OpenEdition special file that defines a null or a
dummy file (for example, /dev/null).

7.2.3 DFSMS/MVS NFS Client Commands

The following commands are provided with NFS client support on the OS/390 and
MVS platforms:
mvslogin Login function to DFSMS/MVS NFS server
mvslogout Logout function from DFSMS/MVS NFS server
showattr Display attributes established for a site or mount point of the
DFSMS/MVS NFS server
nfsstat Display NFS client statistical information, or reset the statistical
information to zero
showmount Display remote NFS server mount information

Chapter 7. Distributed Data Access 99

 os22mvs Convert line delimiter from OS/2 format to MVS format. The
carriage-return line-feed pairs (CR, LF) in the input line are
converted to a newline (NL). No other conversions are performed.
Input and/or output files can be local or remote files.
mvs2os2 Convert line delimiter from MVS format to OS/2 format. The NL in
the input file is converted to CR LF pairs. No other conversions are
performed. Input and/or output files can be local or remote files.

The command programs are intended to run in an OpenEdition MVS shell

environment. They are not implemented as TSO commands. The input/output
file must be an HFS file object. Conventional MVS data sets are not supported.

7.2.4 DFSMS/MVS NFS Server Enhancements

With DFSMS/MVS V1R3, data set attributes such as size and record format are
accessible for migrated SMS-managed data sets, without having to recall the
data set if the data set is migrated under DFSMS/MVS V1R3. Supported
SMS-managed data set types are: ,
• Physical sequential (PS) data set
• VSAM data set
− Entry-sequenced data set (ESDS)
− Key-sequenced data set (KSDS)
− Relative-record data set (RRDS)
− Variable-length relative-record data set
• Partitioned data set extended (PDSE) and partitioned data set (PDS)
members, treating each member as a separate file.

Migrated PDS and PDSE members are not supported because a migrated
member ′s attributes are not accessible through this interface.

7.2.4.1 Obtaining Migrated Data Set Attributes without Recall

The DFSMS/MVS V1R3 NFS server can obtain the attributes of a supported
SMS-managed migrated data set without recalling the data set. To take
advantage of this function, the data set must have been accessed by the
DFSMS/MVS NFS server before migration.

Attributes such as file size and time stamp are saved to DASD. Subsequent file
size requests do not cause a recall of the supported SMS-managed migrated
data set, thus improving performance.

If the data set is modified outside the server by a non-NFS application (for
example, by ISPF edit) before it was migrated, the stored file size could be
incorrect. When the data set is accessed again by the DFSMS/MVS NFS server,
it must be recalled to determine the correct file size.

7.2.4.2 Deleting Migrated Data Sets

When a request is made by an NFS client to delete a supported SMS-managed
migrated data set on the DFSMS/MVS NFS server, the data set will be deleted
without recalling the data set. For PDS and PDSE migrated data sets, the data
set will be recalled to read its member information.

100 DFSMS/MVS V1R4 Technical Guide

7.3 DFSMS/MVS Distributed FileManager
When developing applications in a distributed environment, have you ever
wanted to run your OS/2 programs using the existing host-based data sets,
without having to write additional code on the host side, translate data and
record formats, or transfer the data sets to your workstation? Using the
Distributed FileManager (DFM) set of products, you can access and manipulate
existing host data sets. If the data sets are under the control of an operating
system that supports DFM, all you need to do is write your application, define the
logical connection to your host system, and define the data translation that you
need. Then let DFSMS/MVS DFM do the rest.

DFM as a client/server concept is implemented as a set of products that act as

the interface between your applications and the file data, irrespective of where
the data resides. This implies that the data may be local to the application or
remote to the application on either a LAN server or a mainframe server on a
WAN. This product set comprises client products and mainframe-based
products.

The DFM product set is based on the Distributed Data Management (DDM)
Architecture that has introduced the terms source and target for data requests.
These terms are equivalent to the terms client and server , respectively. One of
the primary ingredients of the Distributed FileManager suite of products is Data
Access Services (DAS) for both record-oriented and stream-oriented data on
local systems and remote servers. With DFM you can invoke commands to
create, delete, rename, and copy data sets on a remote system.

DFSMS/MVS DFM uses the DDM protocol, which enables like and unlike
computer systems to share file systems across a network. It enables your
MVS/ESA system to act as a server (target) to remote client (source) systems.
DFSMS/MVS DFM is designed to work with operating system platforms that
support DDM requests.

DFSMS/MVS DFM enables users on authorized client systems to remotely

access data on MVS/ESA server systems as if the data were local to the users.
Users can remotely perform such tasks as creating, reading, and deleting
MVS/ESA data sets through local commands. Because DDM clients and servers
communicate through common DDM commands, users on the client systems do
not need to use MVS/ESA command language.

DFSMS/MVS DFM accepts access with a record application programming

interface (API) or byte stream access with a byte stream API. DFSMS/MVS DFM
allows client systems to access the following data set types on MVS/ESA server
systems:
• PS data set
• VSAM data set
− ESDS
− KSDS
− RRDS
− Variable-length RRDS
• PDSE and PDS members, treating each member as a separate file.
Note: DFM also supports path processing through an alternate index for VSAM
data sets.

Chapter 7. Distributed Data Access 101

7.3.1 Distributed FileManager DataAgent
The DFM DataAgent, shown in Figure 18, is an extension to the DFSMS/MVS
DFM component of DFSMSdfp and to the DFM component of SMARTdata
UTILITIES that enables remote callers on OS/2, AIX, and Windows NT to invoke
DFM DataAgents to execute specified routines on MVS. The functions that may
be performed using this facility include the execution of TSO and REXX
commands as well as user-written programs. A set of model routines is
provided that illustrates specific uses of this function. The DFM DataAgent
enhancement represents a significant extension of the functionality of the remote
workstation application beyond basic data access.

Figure 18. The DFSMS/MVS DFM DataAgent.

The DFSMS/MVS DFM DataAgent can be called by a remote workstation that

uses SMARTdata UTILITIES to specify the DataAgent routine to be invoked. The
DataAgent routines can be user written, IBM written, or vendor written. Several
sample routines are provided, showing how to write a DataAgent routine in a
high-level language, how to invoke DFSORT, and how to invoke TSO functions
such as TSO CLISTs, REXX EXECs, and TSO commands.

The advantage of the DFSMS/MVS DFM DataAgent approach over some other
client/server packages is that it allows client applications to expand their
function beyond basic data access (read, write, update, and delete) by actually
executing MVS jobs that range from running simple TSO and REXX commands to
fairly elaborate programs.

This ability to initiate jobs on remote MVS/ESA systems provides the basis for
much greater flexibility to tailor access from workstation applications to specific
business requirements. For instance, one DataAgent could be invoked to
preprocess data by making extracts from various files and repositories and
placing the data into a temporary work file. The remote workstation application
could then process the data in the temporary file and, upon completion, another

102 DFSMS/MVS V1R4 Technical Guide

DataAgent could do the postprocessing on the temporary file and put the
updates and changes into the appropriate files and repositories.

The DFSMS/MVS DataAgent provides the ability to initiate jobs on the remote
MVS/ESA platform in a manner similar to remote procedure call. This
represents a significant increase in capability as the workstation application′ s
processing ability is extended beyond basic data access by allowing the actual
initiation of jobs from the remote platform.

The DFSMS/MVS DFM DataAgent requires that the SMARTdata UTILITIES be

installed on the workstation.

7.3.1.1 SMARTdata UTILITIES

The SMARTdata UTILITIES offer a solution to some of the complexities of
developing applications and accessing distributed data in a client/server
environment. The utilities include a record-oriented access method, an interface
to remote server data, data description and conversion services, and data
sorting and manipulation tools. The SMARTdata UTILITIES are embedded in
some popular HLLs (COBOL and PL/I) to simplify client/server application
programming.

The workstation utilities and languages work together with complementary

products on the server system (for example, DFSMS/MVS DFM on the MVS/ESA
platform) to provide transparent access to distributed data.

The SMARTdata UTILITIES are:

• Virtual Storage Access Method (VSAM) Local File System . This utility
provides an API and a workstation local file system that can be used to
create, access, and modify record-oriented files on local (OS/2 or AIX)
systems.
• SMARTdata UTILITIES DFM. This utility, using the same application
programming interface as the local VSAM file system, can be used to create,
access, and modify record-oriented files on remote systems from client OS/2,
AIX, and Windows NT systems. In addition, SMARTdata UTILITIES DFM for
OS/2 provides ″stream access″ to execute OS/2 native commands (for
example, dir and type) directed against MVS files.
• Data Description and Conversion (DD&C). This utility offers data conversion
services that can be used when complex data is exchanged between
different operating systems, programming languages, and international
locations.
• SMARTsort. This utility′s data manipulation tools can help you sort, merge,
extract, and organize your data, including record and stream data, whether it
is local or remote or both.

7.3.1.2 Simplified Application Development

The SMARTdata UTILITIES simplify client/server application development by
offering these features:
• Integration with several HLLs
• Callable application programming
• Transparent access to local and remote data
• A single interface to any kind of record data

Chapter 7. Distributed Data Access 103

 • Server-compatible data management capabilities
• Application portability across several platforms

7.3.1.3 Calling the UTILITIES

To simplify client/server application programming, the SMARTdata UTILITIES are
embedded in several HLLs (currently COBOL and PL/I). You can write a client
application in one of the HLLs that exploit the utilities, and let the HLLs call the
utilities; or your application can call the utilities directly, using the APIs that
come with the utilities. First, you use the administrative functions of the utilities
to define your data environment. Then, whether you call the utilities from an
HLL, the APIs, or both, the utilities handle the complexities of finding, accessing,
converting, and sorting your data.

7.3.1.4 Addressing Distributed Data

The SMARTdata UTILITIES facilitate developing and maintaining client/server
applications that access distributed data, by offering transparent data access and
access to record files.

The data routing and conversion services let you write client applications without
concern for the ___location and format of the server data. You do not have to know
whether the data server is local or remote, you do not have to know how to
address the data formats on the server systems, and you do not have to change
your application if the data is relocated.

The record access utility can help you access record-oriented data, in general
used in mainframe environments, even though workstation data is typically
stream-oriented. You can use exactly the same interface to access record data
anywhere, so you do not have to remember, or learn, how to use several
different data management systems.

7.3.1.5 Client/Server Development

The SMARTdata UTILITIES facilitate integrating workstation applications with
server heritage systems, particularly when used with an associated HLL.

The record access utility and the data sorting tools provide you with new data
management capabilities on the workstation that were only available on the
mainframe in the past. They can help you establish a more complete
development environment on the workstation by letting you test against server
data or simulate server data on the workstation.

The new workstation version of server (mainframe) programming languages,

together with the embedded mainframe-style utilities, can help you run
mainframe applications on a workstation. The mainframe compatibility of the
languages and the utilities give you new delivery and development options:
• You can easily move an existing application from the mainframe to the
workstation without moving your heritage data.
• You can develop an application on the workstation and deliver it on the
mainframe without changing the application.

104 DFSMS/MVS V1R4 Technical Guide

 7.3.1.6 Simplified Systems Management
The SMARTdata UTILITIES simplify client/server systems management by
offering these features:
• Data access independent of data administration
• Remote data access at the record and stream level

7.3.1.7 Independent Applications and Data

The SMARTdata UTILITIES include data access functions, which can be called
from a client application program, and administrative functions, which can be
performed independently of an application. The data access functions let you
write a client application without concern for the ___location or format of your data;
and the administrative functions let you define and change the data environment
without modifying your application. The ability to access and administer data
independently gives you the flexibility to redistribute your applications and data
as your business requirements evolve, without requiring any changes in your
application. Some of the utilities include both data access and administration
functions; others are primarily one or the other.

7.3.1.8 Sharing Distributed Data

The SMARTdata UTILITIES, by providing data access at the record and stream
level, virtually eliminate the need to download and upload entire files. As a
result, using the utilities can reduce the problems and costs associated with
multiple copies of files spread across the client/server environment. Sharing
distributed data, rather then replicating it, can help simplify client/server data
administration tasks, improve data integrity and currency, and reduce the use of
system and network resources.

7.3.1.9 Benefits of the SMARTdata UTILITIES

From an application development perspective, the SMARTdata UTILITIES handle
the complexities of client/server data access and bring the power of mainframe
data management capabilities to the workstation. In addition, the utilities are
embedded in the workstation version of some mainframe programming
languages, thereby simplifying multiplatform application programming.

From a system management perspective, the SMARTdata UTILITIES give you the
flexibility to develop, migrate, and run your applications independently of the way
you distribute your data. By making it easier to share data without replicating it,
the SMARTdata UTILITIES can help simplify data administration, enhance data
integrity, and reduce the overall costs of a client/server environment.

7.3.2 DFM DataAgent Invocation

All of the functions and benefits of the SMARTdata UTILITIES do not alleviate the
need to process data on the mainframe that does not fit in a predefined
transaction and the need to have a way of invoking TSO functions such TSO
CLISTs, REXX EXECs, and TSO commands. This lack of functionality will be
relieved with the DFSMS/MVS DFM DataAgent. A sample DataAgent is provided
to invoke a TSO CLIST, REXX EXECs, or TSO commands.

A DFSMS/MVS DFM DataAgent can also be used to extract data from MVS files
and databases at the beginning of a workstation application in preparation for
subsequent retrieval by the client through normal SMARTdata UTILITIES DFM
interfaces. You could, for example, use DFM DataAgent to access data sets not
otherwise supported by DFM and copy all or some of the data to a temporary file

Chapter 7. Distributed Data Access 105

 that is supported. If the DataAgent returns the name of the temporary file to
DFSMS/MVS DFM, an SMARTdata UTILITIES DFM application could then work
with this temporary file as if it were the real file. The DataAgent could then copy
any changes to the real file or databases when the declaration for the temporary
file name is deleted.

When a DDMOpen API is presented to SMARTdata UTILITIES DFM, the leading

characters in the file name are used to:
• Determine whether the file is local or remote
• Associate the remote file with the specific DDM server

In this way, all client platforms that have SMARTdata UTILITIES DFM installed
can use the DDMOpen remote file request to MVS/ESA or OS/390 to trigger
DataAgent processing on the host.

For both stream and record access, parameter declarations pertain to the object
for which a DCLFIL (declare file) is done. This could be a file, a directory, or a
drive.

Parameter declarations from multiple sessions are treated independently, even if

they are for the same object. For example, if session one defines a DataAgent
for directory A.B, a different session can define a DataAgent for A.B while
session one′s agent is still active. Depending on which session is used to
access the remote data associated with A.B, one or the other agent will be used.

In any case, two different agents cannot run in the same conversation or session
for a given object. Also, no more than 32 DataAgent routines can be active
concurrently on the DFSMS/MVS system and not more than one instance of a
given DataAgent routine can be run at a time if the DFMXLPRM (locate extended
DataAgent) parameter is used.

A DFM DataAgent can only be triggered from a SMARTdata UTILITIES DFM

application. The DFMACALL sample program is distributed with SMARTdata
UTILITIES. It demonstrates the ability to invoke DFM DataAgent functions from C
applications on workstations running SMARTdata UTILITIES.

7.3.2.1 DDMOpen
The current DDMOpen function is used to provide a file name suffix that can be
used to trigger DFM DataAgent processing on MVS.

Use the DDMOpen function as follows:

DDMOpen (MVS_FileName, FileHandle, AccessMethod, AccIntList,
FileShare, EABuf, ........)

The MVS_FileName can contain a parameter suffix as follows:

MVS_FileName <,parm1(value1)<,parm2(value2)<,.....>>>

where MVS_FileName and parm1 through parmn are passed unchanged to

DFSMS/MVS DFM for processing.

106 DFSMS/MVS V1R4 Technical Guide

7.3.2.2 DDMClose
The current DDMClose function is used to terminate agent processing. It issues
the DDM commands, CLOSE and DELDCL (Delete Declaration), which actually
terminate agent processing by invoking the exit with the DELDCL code point in
the parameter list, if requested.

7.3.2.3 File Name Suffix Parameters

The new DataAgent file name suffix parameters currently recognized by MVS
are:
• AGENT(agent_name<,procedure_parameter>) - Invokes the indicated agent
when a file is declared (at DDMOpen) and, optionally, when the file
declaration is deleted (at DDMClose). The agent_name must specify the
name of a member in SYS1.PROCLIB. Parameters can optionally be
provided for symbolic substitution in the PROCLIB member.
However, allocation will run under the authorization assigned to started
tasks. The agent (running under the user′s authorization) may have to use
dynamic allocation to allocate files that cannot be allocated by started tasks.
Because the agent begins as a started task, unless DFM00 (PROCLIB
member) specifies RESTRICT_START(NO), the first three characters of the
procedure member name must be DFM.
The maximum length of the agent name and its parameters is 107 bytes.
Each parameter in the list of procedure parameters is subject to the MVS
limit of 44 bytes.
The agent runs synchronously. If the PROCLIB member has multiple steps,
any file name changes or return code settings will be propagated to the later
steps and will be returned to DFM only after the last step is executed.
• PARM(agent_parameter_list) - Parameters to be passed to the agent routine
when it begins running in the DataAgent address space. Parameters are
converted to upper case and concatenated to any parameters provided in the
PROCLIB member.
If PARM is specified, the PROCLIB member must contain the JCL statements
as provided by the sample DFMX0001, which specifies the DFMINIT
parameters so as to run a DFSMS/MVS DFM DataAgent address space
initialization routine as the first program in the address space. This causes
DFSMS/MVS DFM to pass a supplementary run-time parameter list to the
DataAgent routine to allow the routine to return an error code and additional
reason codes to DFM. The agent parameter list specified is concatenated
with DFMINIT before the DataAgent routine is invoked.
This parameter is ignored if AGENT is omitted.
• PGM(prgram_name) - Name of program (DataAgent routine) that DFM should
invoke after initialization. If omitted, the program invoked will default to the
agent name requiring that you have identically named MVS load module and
PROCLIB members.
• START(job_name<,job_parameters>) - Name of PROCLIB member
representing a job or procedure to be started asynchronously. Unless
DFM00 specifies RESTRICT_START(NO), the first three characters of the
procedure or command must be DFM.
Optional parameters can also be provided. The MVS limit for the total length
of the job name and its parameters is 124 bytes.

Chapter 7. Distributed Data Access 107

 DFSMS/MVS DFM verifies that an address space for running the procedure
was created but does not verify that the procedure exists or that it ever
completes successfully; that is, the started job runs asynchronously.

It is possible to run some existing PROCLIB members that may not have
particular initialization requirements by using only the AGENT keyword, we do
not recommend that you do so because return codes will not be passed back to
DFM. Also, there will usually be a need for extended parameter passing. For
these two reasons, use the AGENT parameter in conjunction with the PARM
and/or PGM parameter, even if the PARM parameter is the value of PARM() or
the PGM name is the same as the agent name.

DFSMS/MVS DFM imposes a limit of 255 bytes for the file name and file name
suffix and therefore for the total length of the parameter (AGENT,
PARM,PC_CCSID,START) that can be passed.

7.3.3 DataAgent Processing (DFM Target)

The DataAgent is first called when a stream or record file is declared with a file
name suffix specifying the agent name. For example, assume the remote file
name sent to MVS is ″A.B,AGENT(DFMXUTIL)″. The suffix specifying the agent
name is AGENT(DFMUXITL).

The comma, which is invalid as an MVS file name character, is used to

distinguish the MVS file name from its suffix. If the file name sent to MVS is
″A.B.C,PARM1(xxx),PARM2(yyy)″, the comma is used to distinguish the MVS
parameters from each other as well as the file name.

On the client system, a sample SMARTdata UTILITIES application called

DFMACALL is available. On an MVS system you are responsible for ensuring
that the DataAgent name is unique in SYS1.PROCLIB and in its load library. The
prefix DFM has been reserved for DFSMS/MVS DataAgent names to help ensure
uniqueness. Sample DFM agents use the prefix DFM or DFMX and can be
modified and/or renamed as appropriate.

If the DataAgent routine does not exist or an error occurs while loading a
DataAgent, a VALNSPRM (parameter value not supported) will be returned. If
the agent returns an error as a result of a DCLFIL or later call, an invalid request
error reply message (INVRQSRM) will be returned.

7.3.3.1 DataAgent Interface

This discussion of DataAgent processing assumes that the PARM keyword was
specified in conjunction with the AGENT parameter. Otherwise, processing is
triggered once at DCLFIL (declare file) time, and processing is the same as for
any other synchronous, started task.

The DataAgent can be called a number of times. It can be called at DCLFIL time
for preprocessing necessary to build a temporary file for use by an application or
for one-time processing. It can also be called at DELDCL (delete declaration)
time for postprocessing necessary to move temporary file data to its final
destination. Note that this temporary file is located on the server site, so there
is no data traffic over the network.

A second parameter is provided to allow DataAgent routines to request

postprocessing or additional return and/or reason code processing.

108 DFSMS/MVS V1R4 Technical Guide

See sample agent DFMXSORT for a DataAgent routine that does both
preprocessing and postprocessing.

The exit to the DataAgent routine has the following interface:

• Program status: AMODE(31), RMODE(ANY), KEY(8), TCB mode, primary
mode, running in its own address space. The program runs either
APF-authorized or non-APF-authorized, as appropriate.
• Register: Standard MVS register conventions are followed.
Register 1 points to a pointer parameter list consisting of a halfword length
field followed by the character string from the workstation′s PARM field. In
addition, immediately following the first parameter pointer, a 4-byte pointer
to a structure can be used by the DataAgent routine to request that it be
called again, to provide additional error feedback, or to allow the routine to
change the file name.
Existing applications are likely to run unchanged because they will be
unaware of the second parameter pointer. However, it should be fairly easy
to write a new DataAgent application that can access the additional
parameter list, either by looking for a second parameter list pointer after the
first or by using the DFMXLPRM function described in “DataAgent Locate
Extended Parameter List (DFMXLPRM)” on page 110.

DataAgent Parameter List: At DCLFIL and DELDCL time, register 1 points to a

fullword address of a 2-byte length field and a variable length character string
representing the PARM parameter as specified at DCLFIL time. Immediately
following the pointer to the first parameter is a pointer to the DataAgent
Extended Parameter List.

DataAgent Extended Parameter List: DCLFIL and DELDCL processing calls the
DataAgent routine with a second parameter consisting of an area of storage
containing the following:
1. Halfword length field
The length of the parameter list that follows
2. Reserved halfword
Reserved for future use
3. Two-byte command code point
Refer to the DDM architecture for the definitions of code points. In the first
release the agent will only be taken for DCLFIL, code point X′102C′, and
DELDCL, code point X′102D′. This code point can be used by the DataAgent
routine to determine whether to do preprocessing, postprocessing, or
one-time processing.
4. Two-byte code point representing the type of object to be declared
The agent can be taken for an object whose type is DRCNAM (directory
name), code point X′1165′, and for FILNAM (file name), code point X′110E′.
5. Original MVS file or directory name
Two-byte length field and a 54-byte character string containing the file or
directory name padded with blanks. This is the name as provided by the
workstation.
6. Modified MVS file name

Chapter 7. Distributed Data Access 109

 Two-byte length field and a 54-byte character string containing a file name
padded with blanks.
The DataAgent is allowed to return a new length and file name here for
DFSMS/MVS DFM processing using the altered file name (attempts to
change directory names are ignored).
At DELDCL time this field will contain the file name as modified by any
previous DataAgent processing.
7. Eight-byte extended reason code area (for optional use by the DataAgent if
setting register 15 to nonzero).

DataAgent Locate Extended Parameter List (DFMXLPRM): A new callable

system service is provided to allow DFM DataAgent routines written in an HLL to
locate the DataAgent parameter list.

Function syntax:
DFMXLPRM(AGENTNAME,PARMPTR)
− AGENTNAME (input)
This variable contains the name of the agent routine (which is not
necessarily the same as the procedure name) whose extended
parameter list is to be located. It is eight characters long and padded
with blanks, if necessary.
− PARMPTR (output)
A 4-byte pointer to the extended parameter list. Set to zero if the
parameter list cannot be found.

For example, a DataAgent exit written in C could call this function and test the
result as follows:
DFMXLPRM(″DFMXAGENT″,&p_extra);
if (p_extra == NULL) {

The DFMXAGNT sample DataAgent routine is distributed in source form in

SYS1.SAMPLIB, member DFMXAGNT, with the DFSMS/MVS DFM DataAgent.

Other DataAgent Routine Considerations: By default DataAgent routines are run

only once, when the file or directory is declared. However, if a DataAgent
routine wants to do any further processing for the request (for example, if it
needs to do some processing at delete declaration time) it can set the first word
of the first word of the extended reason code area to -1 before returning with
register 15 set to zero.

A nonzero value in register 15 will terminate any further processing for the
DataAgent routine regardless of the extended reason code setting.

If the DataAgent routine is running in an HLL environment (in which it is not

possible to access the second parameter directly), DFMXLPRM can be used to
return a pointer to the extended parameter list. To use DFMXLPRM
SYS1.CSSLIB must be specified in the SYSLIB DD statement concatenation when
the DataAgent routine is link edited.

110 DFSMS/MVS V1R4 Technical Guide

7.3.3.2 Sample DataAgent Routines
Sample DataAgent routines are provided with the DFSMS/MVS DFM DataAgent
function. These routines can be run as is, or they can be used as models to
create your own routines.

PROCLIB Sample DFMX0001: DFMX0001 is a sample PROCLIB member that can

be used as a template for setting up PROCLIB members required by DataAgent
routines. It is set up to discard the JOBLOG listing that is produced. If you do
not discard the output, you may have to issue the ′$ps′ command periodically.

The general procedure for converting an old PROCLIB member so that it can run
as a DataAgent is to:
1. Copy the member to a new PROCLIB member (preferably with a prefix of
DFMX).
2. Add DFMINIT = to the PROC statement.
3. Change the program name on the EXEC statement to PGM=&DFMINIT
4. Invoke the DataAgent routine by specifying AGENT(new_procname)
PGM(original_program_name).

DataAgent for Full TSO Support: DFMXTSO is provided as a general-purpose

TSO agent. It is intended to be used with the IKJEFT01 function so its
restrictions are as documented by TSO for batch execution. The agent can be,
for example, invoked from an OS/2 command file or an SMARTdata UTILITIES
DFM application to force execution of a CLIST or REXX EXEC on an MVS/ESA or
OS/390 system.

IKJEFT01 will obtain commands to issue from both the parameter list and the
userid.DFMXTSO.SYSTIN file. Output will be written to the
userid.DFMXTSO.SYSTSPRT file.

SYSTPRT can be browsed from the workstation to determine the results of the
commands executed.

DataAgent for Quick TSO Support: DFMQTSO is provided as a general purpose

TSO agent. It links to the IKJTSOEV function so its restrictions are as
documented by TSO (for example, it is restricted to nonauthorized commands
and programs). The agent can be, for example, invoked from an OS/2 command
file or an SMARTdata UTILITIES DFM application to force execution of a CLIST or
REXX EXEC.

DFMQTSO is available as a source file in SYS1.SAMPLIB.

The agent routine is written in such a way as to obtain commands to issue from
the parameter list and not from a SYSTIN file. Output is written to
userid.DFMQTSO.SYSTSPRT.

SYSTSPRT can be browsed from the workstation to determine the result of the
command executed. Remember that the SYSTSPRT file may be empty if the
PARM field specifies an invalid command.

SMARTdata UTILITIES-Based MVS DFM DataAgent Caller: A sample

SMARTdata UTILITIES application called DFMACALL is available as a sample C
program, which facilitates invoking TSO CLISTs and DataAgent routines from the
workstation.

Chapter 7. Distributed Data Access 111

 The application will accept an assigned drive letter and a file name containing
the TSO command input. For example:
DFMACALL QTSO driveletter: TSOcommandline <DISPLAY>
DFMACALL TSO driveletter: <TSOcommandline> <DISPLAY>
DFMACALL AGENT driveletter: <remote_filename> MVS procedure<,proc_parm>
<PGM program_name> <PARM program_paramaters> <DISPLAY>
DFMACALL START driveletter: MVS procedure<procedural_parms>
DFMACALL driveletter: remote_filename<,filename_suffix> <DISPLAY>

Note that the remote file name is optional for AGENT, disallowed for QTSO, TSO,
and START, and required for the free-form invocation. If a remote file name is
specified, it must exist but other than checking that it can be opened for input, it
is not used by DFMACALL itself.

QTSO invokes quick TSO exit DFMQTSO, TSO invokes regular TSO exit
DFMXTSO, AGENT invokes the specified MVS procedure, START runs the
specified MVS command, and the last is the free form getting its parameters
from the filename_suffix, if any.

To use this application you have to initialize the procedure′s input file on MVS.
The provided samples use userid.AGENT_NAME.SYSIN or
userid.AGENT_NAME.SYSTSIN, depending on whether the agent is DFMXTSO.
For example, when you use the TSO DataAgent routine DFMXTSO, the SYSTSIN
input file might contain something like %MYCLIST PARM1 ....PARMn.

After running the application, you can view the results in the MVS SYSOUT file,
for example, in userid.AGENT_NAME.SYSPRINT (or
userid.AGENT_NAME.SYSTSPRT for TSO agents DFMQTSO and DFMXTSO).

Because TSO batch defaults to no prefixing of data set names, you might want to
specify ″profile prefix(userid)″ in either the PARM or SYSTSIN.

Other MVS DFM Sample DataAgent Routines: Sample agent DFMXAGNT,

written in C, provides a starting point for user- or vendor-written routines and
demonstrates that HLLs can be used to write DataAgent routines.

Sample agent DFMXSORT sorts an input file based on column numbers specified
in the parameter list, builds a temporary file consisting of the sorted output, and
returns the name of the temporary file to DFM for subsequent retrieval by
SMARTdata UTILITIES DFM.

7.3.3.3 START Command Consideration

By default, the start function, START(job_name), will ensure the job_name (or
procedure name) starts with DFM before issuing the start command. Because
some customer might not have the appropriate RACF (or other vendor security
product) definitions defined to tolerate certain commands or procedures, the
function defaults to only starting jobs or procedures whose prefix is DFM.

The system PARMLIB DFM00 member contains a new DFSMS/MVS DFM

parameter, Start command control. If this parameter is set to YES, DFSMS/MVS
DFM AGENT and START command support is limited to those jobs or procedures
with names that are at least four characters long and that begin with DFM. The
variable RESTRICT_START defaults to YES.

112 DFSMS/MVS V1R4 Technical Guide

7.3.3.4 Installation Considerations
You will have to determine the appropriate setting for the RESTRICT_START and
run the installation jobs for the sample applications. You may also have to
preallocate some input or output files for TSO processing.

Because DataAgent routines can initiate long-running processes, Advanced

Program-to-Program Communication (APPC) message limit and timeout settings
should be adjusted accordingly.

7.3.3.5 DFSMS/MVS System Messages

The new system message issued by DFSMS/MVS DFM is:

GDE007E module ERROR INVOKING FUNCTION:function return_code

reason_code

Explanation: During DFM DataAgent processing, the MVS function shown failed
with the indicated return and reason codes.

However, common errors have more specific text. For example, if the DataAgent
routine cannot be found in JOBLIB, STEPLIB, or LPALIB, ″LOCATING MODULE
DataAgent_routine_name″ will be substituted for function, return_code, and
reason_code.

System Action: DFM DataAgent processing is terminated.

Operator Response: Notify your System Programmer.

System Programmer Response: Refer to documentation of the indicated function

to determine the meaning of the return and reason codes. Refer to the job log
for any related console messages that may have occurred at the time of error.

Descriptor Code: 4 (informal system status)

Routing Information: 10 (system/error maintenance)

Chapter 7. Distributed Data Access 113

114 DFSMS/MVS V1R4 Technical Guide
Chapter 8. DFSMS/MVS Optimizer

The DFSMS Optimizer is not part of the DFSMS/MVS V1R4 Preview

announcement; it was announced in November 1994. With DFSMS/MVS V1R4 the
Optimizer will be a priced feature, and the code will be shipped with the base
product. The Optimizer code has been totally refreshed and it provides support
for OS/390 in addition to MVS/DFP Version 3 Release 3.

In this chapter we review the main features of the DFSMS Optimizer.

The DFSMS Optimizer uses historical and real-time data to provide an overall
picture of data usage on each system. It gives an installation the ability to
understand how it is managing storage today. With that information the
installation can then make informed decisions about how it should manage this
storage in the future.

The Optimizer uses the following analyzers to provide detailed and summarized
information about storage use (Figure 19 on page 116):
• Performance analyzer
• Management class analyzer

The DFSMS Optimizer uses input data from several sources and processes it,
using an extract program that merges the data and builds the Optimizer
database.

By specifying different filters, you can produce reports that help you build a
detailed management picture of your enterprise. You can use the charting
facility to produce color charts and graphs from the report data to greatly
enhance the value of this information.

In addition, the HSM Monitor/Tuner is available to keep you informed about the
status of DFSMShsm′s activity and provides dynamic control and tuning
capability.

By utilizing a historical database, the Optimizer can provide information about

how the system was used previously and project how it may be used in the
future. Real-time data can provide information about how the system is currently
being used. The Optimizer provides what-if simulations for performance,
capacity planning, system tuning, and the application of SMS technology, both
hardware and software.

8.1.1 Input Data for the Optimizer

As shown in Figure 19 on page 116, the Optimizer uses data from Resource
Measurement Facility (RMF), Cache RMF Reporter (CRR), system management
facility (SMF) (including HSM information), and SMS to help an installation
develop a picture of how storage is being managed. The Optimizer builds its
database by correlating data from different record types to produce a complete
picture of the data processing environment. All of these records are required in
order to do that. The data kept in the DFSMSopt database is usually 85-90%
reduced from the raw SMF data.

 Copyright IBM Corp. 1997 115

Figure 19. Elements of the DFSMS/MVS Optimizer

8.1.1.1 Data Extract

The required records are as follows:
• SMF records
− OPEN/CLOSE records (type 14, 15, 62, and 64)
− STEP/JOB end (type 30, subtypes 4 and 5)
− Delete/rename (types 17, 18, 65, 66)
− Data set performance (type 42, subtypes 5 and 6)
− Tape demount (type 21)
• RMF record types 70 and 74
• Cache RMF Reporter - record 245 (default)
• HSM Functional Statistic Records (FSR) - record 241 (default)

8.1.2 Performance Analyzer

The performance analyzer allows an installation to analyze performance data of
the subsystem, volumes, and data sets. Using the data from this analyzer, an
installation can determine the causes of performance problems such as device
skew, I/O constraints, or I/O-intensive periods. With this information, an
installation can make informed decisions about capacity planning, isolate I/O
performance problems, and perform what-if simulations for all subsystems in the
sysplex.

The performance analyzer helps with:

• What-if simulations - identifying the effects of various changes, such as

116 DFSMS/MVS V1R4 Technical Guide

 − Changing the size of cache storage
− CPU hardware compression
− Data set striping
− Data in memory placement
− Conversion to extended format data sets
• Current state analysis, including
− Batch window analysis
− Reporting on average holding time and cache hit ratios
− Sorting top data sets by
- I/O rate
- Megabytes transferred
- I/O response time
- I/O intensity
• Performance cost components
− Daily megabyte cost of cache and nonvolatile (NVS) storage
− Daily megabyte cost of expanded storage

8.1.3 Management Class Analyzer

The management class analyzer provides cost-benefit analyses and what-if
simulations of management class policies for SMS and non-SMS data. With this
analyzer the storage costs and data access patterns are used as input when a
report is run.

The management class analyzer can also be used to predict the amount of level
0 (L0) and migration level (ML1) storage that would be required if a new set of
migration attributes is used.

Additionally, the management class analyzer can be used to determine the most
cost-effective management class assignments for data sets that are to be
converted to SMS or to tune the current management class settings to reduce
possible DFSMShsm thrashing.

8.1.4 Charting Facility

The charting facility is an OS/2 Presentation Manager program that provides an
easy-to-use graphical user interface to download files, view and print color
charts, and create slide shows.

While viewing each chart, you can dynamically set attributes—style of chart,
colors, text placement—to create the charts that best fit the needs of your
installation.

The charting facility accepts any of the report files and automatically converts
each file into multiple charts. Color charts are available in various formats
including column, tabular, and three dimension.

8.1.5 HSM Monitor/Tuner

Real-time data is the most current data activity on a system. To identify
real-time data, an installation must monitor data and system resources. The
HSM Monitor/Tuner is a client/server solution that provides real-time monitoring
of all major HSM functions on multiple systems.

Chapter 8. DFSMS/MVS Optimizer 117

 In addition, the HSM Monitor/Tuner can automatically monitor and tune
DFSMShsm through the use of more than 70 trigger events (20 of which are
threshold-based) that invoke OS/2 REXX EXECs. These EXECs can be
customized to issue commands directly to the host that turned on the trigger
event, issue NetView alerts, or perform automated responses to the event.

The HSM Monitor/Tuner OS/2 interface enables all DFSMShsm functions running
on each of the associated MVS hosts. It helps you set up system resources
correctly to ensure that HSM runs in an efficient and timely manner.

118 DFSMS/MVS V1R4 Technical Guide

Chapter 9. NaviQuest and Catalog Search Interface

In this chapter we provide an overview of NaviQuest and the Catalog Search

Interface (CSI) capabilities. The NaviQuest component of DFSMSdfp assists
customers in their initial migration to system-managed storage (SMS) and their
ongoing configuration testing assurance and maintenance.

NaviQuest enables customers to perform many storage-related tasks in batch, in

support of an SMS environment. Batch support includes:
• Selected DFSMS configuration creation and modification
• Testing - performing regression on changes to ACS routines and the DFSMS
configuration
• Translation of ACS routines
• Validation of the DFSMS configuration
• Storage reporting
• Additional automation for ongoing, routine storage tasks

CSI was developed to give users easy read-only access to information stored in
ICF catalogs. It provides an alternative to using the AMS LISTCAT command.

9.1 NaviQuest
SMS provides comprehensive and cost-effective management of an installation′ s
most valuable asset, data. Recognizing that some customers have not yet
exploited SMS through the DFSMS/MVS product, additional assistance is being
provided through the NaviQuest component of DFSMSdfp.

The NaviQuest component was introduced in DFSMSdfp as an SPE in

DFSMS/MVS V1R3.

NaviQuest helps customers who are implementing SMS for the first time. It
enables the storage administrator to test policies and configurations before
running production data. NaviQuest also helps customers who have already
implemented SMS and want to restructure their SMS routines to add new data
types to be managed or exploit their SMS investment by using new or additional
SMS functions.

NaviQuest offers solutions for simplified DFSMS testing and additional storage
management capabilities:
• Familiar ISPF/ISMF panel interface
• SMS implementation assistance
• Fast, easy, bulk test case creation
• Automatic class selection (ACS) testing automation (that is, regression
testing)
• Storage reporting assistance
• Additional tools to aid with storage administration tasks.

NaviQuest provides batch execution of several key SMS functions, including:

• Creation of data set and volume listings
• ACS routine translation

 Copyright IBM Corp. 1997 119

 • ACS routine validation

The DFSMSdfp NaviQuest component supports the DFSMS Fast Implementation

Techniques (DFSMS FIT) methodology but is not restricted to being used with
this methodology. DFSMS FIT is documented in these ITSO redbooks:
• Get DFSMS FIT (SG24-2568)
• DFSMS FIT Installation Examples (SG24-2569)
• DFSMS FIT Forms and Foils (SG24-2570)
• DFSMS FIT Process Guide (SG24-4478)
Both NaviQuest and DFSMS FIT were developed in conjunction with actual
customer system-managed storage implementations.

9.1.1 Enchanced ACS Management Option Panels

In this section we review some of the enhanced ACS Management Option Panels
that the NaviQuest component brings in support of SMS. For a more detailed
description of all of the fields on each panel, refer to the DFSMS/MVS Version 1
Release 4 NaviQuest User ′ s Guide (SC26-7194).

Rather than presenting every panel and discussing every field, in this section we
show the main panels as they apply to two main areas for which the NaviQuest
component is likely to be used:
• Regression testing for changes to ACS routines and the DFSMS configuration
• Interactive selection of storage functions for batch processing

9.1.1.1 Regression Testing Using NaviQuest

The sequence of panels we present in this section shows how NaviQuest can be
used to identify any changes to data class, storage class, management class,
and storage group assignments, for a particular grouping of data sets from a
data type (for example, TSO, batch production, and test), referred to as a data
subtype , after an update to the ACS routines.

Your installation probably has an ISMF Saved List that contains the names and
attributes of many data sets that are directly associated with TSO. The ISMF
Saved List is created from ISMF′s Data Set Selection Entry Panel and then on
the resulting Data Set List panel by typing SAVE TSOLIST on the command line.
When using NaviQuest, it would be normal practice to limit the ISMF Saved List
to a single grouping of data sets (or data subtype) that have the same data
class, storage class, management class, and storage group policy names.

On the ISMF Primary Option Panel shown in Figure 20 on page 121, choose
option 11, Enchanced ACS Management, to go directly to the NaviQuest Primary
Option Menu (POM), which includes seven NaviQuest options, plus the option to
Exit (see Figure 21 on page 121).

120 DFSMS/MVS V1R4 Technical Guide


Panel Help

--------------------------------------------------------------------------------
DGTSMMD2 ISMF PRIMARY OPTION MENU - DFSMS/MVS 1.4
Enter Selection or Command ===> 11_____________________________________________

Select one of the following options and press Enter:

0 ISMF Profile - Change ISMF user profile

1 Data Set - Perform Functions Against Data Sets
2 Volume - Perform Functions Against Volumes
3 Management Class - Specify Data Set Backup and Migration Criteria
4 Data Class - Specify Data Set Allocation Parameters
5 Storage Class - Specify Data Set Performance and Availability
6 Storage Group - Specify Volume Names and Free Space Thresholds
7 Automatic Class Selection - Specify ACS Routines and Test Criteria
8 Control Data Set - Specify System Names and Default Criteria
9 Aggregate Group - Specify Data Set Recovery Parameters
10 Library Management - Specify Library and Drive Configuration
11 Enhanced ACS Management - Perform Enhanced Test/Configuration Management
C Data Collection - Process Data Collection Function
L List - Perform Functions Against Saved ISMF Lists
R Removable Media Manager - Perform Functions Against Removable Media
X Exit - Terminate ISMF
Use HELP Command for Help; Use END Command or X to Exit.
 
Figure 20. ISMF Primary Option M e n u

Panel Help

--------------------------------------------------------------------------------
ACBSMDP0 ENHANCED ACS MANAGEMENT - NAVIQUEST PRIMARY OPTION MENU
Enter Selection or Command ===> 1______________________________________________

Select one of the following options and press Enter:

1 Test Case Generation

2 ACS Test Listings Comparison
3 Enhanced ACS Test Listing
4 Test Case Update with Test Results
5 SMS Report Generation
6 Model Commands Generation
7 Batch Testing/Configuration Management
X Exit

Use HELP Command for Help; Use END Command or X to Exit.

 
Figure 21. DFSMSdfp NaviQuest Component Primary Option M e n u

Choose option 1, Test Case Generation, to get to the Test Case Generation
Selection Menu, which, as you can see in Figure 22 on page 122, enables you to
create bulk test cases from one of four different sources of data.

Chapter 9. NaviQuest and Catalog Search Interface 121

 Option 3, SMF Data, assumes that the IGDACSSC storage class exit, provided by
the CBIPO, which creates and stores SMF Type 127 records and stores them in
the SMF data set, is already installed. The IGDACSSC storage class exit is
invoked every time a data set is allocated. Hence the data collected reflects
only new data sets allocated since the SMF data set itself was defined. Choose
option 3 to invoke the ACSTST program, also provided by the CBIPO, to process
these Type 127 records to generate the test cases.

You would choose option 4, VMA Extract Data, to generate test cases for tape
data sets and tape volumes that should be stored and accessed from a
system-managed tape library (that is, the IBM 3494 and/or 3495 Automated Tape
Library Data Servers).

In the example in Figure 22, we use option 1, Saved ISMF List. We call the list
TSOLIST .

Panel Help

--------------------------------------------------------------------------------
ACBSFLG4 TEST CASE GENERATION SELECTION MENU
Enter Selection or Command ===>1_______________________________________________

Select the input data to be used and press Enter:

1 Saved ISMF List

2 DCOLLECT Data
3 SMF Data
4 VMA Extract Data

Use HELP Command for Help; Use END Command or X to Exit.

 
Figure 22. Test Case Generation Selection M e n u

Figure 23 on page 123 shows the Test Case Generator from Saved ISMF List
Entry Panel from which test cases can be generated from a Saved ISMF List. In
this example, the DFSMS policies are assigned on the basis of the information
contained in the Saved ISMF List, which contains such information as data set
name, data set organization, the disk volume on which the data set was stored
on, and the size of the data set.

It is also possible to specify other variables on this panel that would apply to all
test cases in the Saved ISMF List, such as ddname and program name.

We continue to use the name of the previously Saved ISMF List: TSOLIST. Each
test case is stored as a separate member in a library; in this example, the
library is ′NAVIQ.ONL.TESTLIB.′ Each test case is given a member name (for
example, TSOL1, TSOL2, ...).

122 DFSMS/MVS V1R4 Technical Guide


Panel Help

--------------------------------------------------------------------------------
ACBDFLG1 TEST CASE GENERATOR FROM SAVED ISMF LIST ENTRY PANEL
Command ===> __________________________________________________________________

To generate test cases, specify the following information and press Enter:
Saved ISMF List . . . . . . TSOLIST_ (Data set list)

Member Name Prefix . . . . TSOL (1 to 4 alpha characters)

Test Case PDS . . . . . . . ′ NAVIQ.ONL.TESTLIB′__________________________
Replace Existing Prefix . N (Y or N)

ACS Test Case Variables:

More: +
Applic . . . . ________ DD . . . . . . ________
Def_dataclas . ________ Def_mgmtclas . ________
Def_storclas . ________ Filenum . . . ____
Group . . . . . ________ Job . . . . . ________
Label . . . . . ________ Libname . . . ________
Mvsgp . . . . . ________ Pgm . . . . . ________
Retpd . . . . . ________ Storgrp . . . ________

Use DOWN Command to Scroll Forward; Use UP Command to Scroll Backward:

Use HELP Command for Help; Use END Command to Exit.
 
Figure 23. Test Case Generator from Saved ISMF List Entry Panel

After these test cases have been generated, if a typical member is browsed, it
would contain information such as:

BROWSE NAVIQ.ONL.TESTLIB(TSOL1)
DESCRIPTION1:
TEST CASE CREATED 96/11/04 at 19:58:5 BY LITTLEP
DSN: NAVT001.DAILY.REPORT
DSORG: PS
DSTYPE: PERM
ACSENVIR: RECALL
JOB: DFHSM
SIZE: 55
STORCLAS: STD
MGMTCLAS: STD
NVOL: 1
VOL: 01
MIGRAT
UNIT: 3390

Once the test cases have been generated, you can test them, using option 7,
Automatic Class Selection (see Figure 20 on page 121). Save the output from
this run in the LISTING data set as it will act as the baseline against which future
runs with modified ACS routines are checked.

The full procedure for running regression testing is documented in the NaviQuest
User ′ s Guide (SC26-7194). Basically whenever the ACS routines are to be
changed such that a new data subtype is to be managed, you must do the
following:
1. Ensure that the test case library contains the new data subtype.
2. Test these test cases against the new ACS routine, using option 7, Automatic
Class Selection, and write the output to a new listing file.

Chapter 9. NaviQuest and Catalog Search Interface 123

 3. Use NaviQuest to perform the regression test as illustrated in Figure 24 on
page 124.

To perform the comparison, choose option 2, ACS Test Listings Comparison,

from the NaviQuest POM (Figure 21 on page 121), to bring up the ACS Test
Listings Comparison Panel as shown in Figure 24.

This panel contains the names of the listing files entered, from before and after
the changes were made to the ACS routines, the name of the library used to
hold the test cases, and the names of two output data sets:
1. Comparison Results Data Set, a sequential report file used to highlight the
differences for the exception cases
2. Exception Test Case PDS, a data set that holds copies of the members that
were the exception test cases.
The Exception Test Case PDS also holds details of all test cases that either have
never been run before or have been given different data class, storage class,
management class, or storage group assignments. If these exceptions identify
errors in assigning the DFSMS policies, then the ACS routines must be recoded
and the process rerun.
Note: The only valid exceptions should be for the new data subtype that has
never been through this process before and for which, therefore, expected
results are not stored in the test case library for the new test cases.

Panel Help

--------------------------------------------------------------------------------
ACBDFLC1 ACS TEST LISTINGS COMPARISON PANEL
Command ===> __________________________________________________________________

To compare ACS listings, specify the following information and press Enter:
Input Data Sets:
Base ACS Test Listing (Before latest ACS routine changes)
===> LISTING_________________________________________________
New ACS Test Listing (After latest ACS routine changes)
===> NEW.LISTING_____________________________________________
Reference Data Set for Compare:
Test Case PDS (Test source for listings above)
===> ′ NAVIQ.ONL.TESTLIB′_____________________________________

Output Data Sets:

Comparison Results Data Set (Summary of exception test cases)
===> COMPARE.REPORT__________________________________________
Replace Contents if DSN Exists . . Y (Y or N)
Exception Test Case PDS (Contents of exception test cases)
===> EXCEPTNS________________________________________________
Replace Contents if DSN Exists . . Y (Y or N)

Use HELP Command for Help; Use END Command to Exit.

 
Figure 24. ACS Test Listings Comparison Panel

If the ACS routines have logic errors that produce incorrect assignments for any
of the data types (or data subtypes), you can use the Enhanced ACS Test Listing
Entry Panel (select option 3, Enchanced ACS Test Listing, on the NaviQuest
POM) to find further information for each data set that was tested. As you can
see in Figure 25 on page 125, you can choose which additional information
should be included in the output listing.

124 DFSMS/MVS V1R4 Technical Guide


Panel Help

--------------------------------------------------------------------------------
ACBDFLX1 ENHANCED ACS TEST LISTING ENTRY PANEL
Command ===>___________________________________________________________________

To generate enhanced ACS test listing, specify the following and press Enter:
ACS Test Listing
===> NEW.LISTING_____________________________________________

Enhanced ACS Test Listing

===> ENHANCED.LISTING________________________________________
Replace Contents if DSN Exists . . Y (Y or N)

Fields to Include in Enhanced ACS Test Listing: (Y or N)

DSN . . . . . Y JOBNAME . . . . Y
EXPDT . . . . N SIZE . . . . . . N
UNIT . . . . . Y PROGRAM . . . . Y

Use HELP Command for Help; Use END Command to Exit.

 
Figure 25. Enhanced ACS Test Listing Entry Panel

Once you have corrected all of the ACS routine errors and ensured that all
exceptions are valid (that is, only for the new data subtypes), you can use
NaviQuest to place the results of the test into the test case definitions, as the
saved expected results for later regression testing.

To save the test results, choose option 4, Test Case Update with Test Results,
from the NaviQuest POM to get the Test Case Update with Test Results Entry
Panel. Figure 26 on page 126 shows this panel filled in with the:
1. Latest listing (NEW.LISTING) after the changes to the ACS routines
2. Names of the comparison results and exception test case data sets
3. Test case library

The test case members for the exceptions are read and copied into the test bed
library. The saved expected results are obtained from the comparison report
and are also saved in the test bed library.

Chapter 9. NaviQuest and Catalog Search Interface 125


Panel Help

--------------------------------------------------------------------------------
ACBDFLU1 TEST CASE UPDATE WITH TEST RESULTS ENTRY PANEL
Command ===>___________________________________________________________________

To update test cases with test results, specify the following and press Enter:

Input Data Sets:

New ACS Test Listing (After latest ACS routine changes)
===> NEW.LISTING_____________________________________________
Comparison Results Data Set (Summary of exception test cases)
===> COMPARE.REPORT__________________________________________
Exception Test Case PDS (Contents of exception test cases)
===> EXCEPTN_______________________________________

Input/Output Data Set:

Test Case PDS
===> ′ NAVIQ.ONL.TESTLIB′___________________________

Use HELP Command for Help; Use END Command to Exit.

 
Figure 26. Test Case Update with Test Results Entry Panel

9.1.1.2 Running Storage Administration Functions in Batch

Running storage administration functions in batch results in a huge productivity
improvement. From NaviQuest POM, select option 7, Batch
Testing/Configuration Management (see Figure 21 on page 121), to get to the
Batch Testing/Configuration Selection Menu (Figure 27 on page 127), where you
can choose an appropriate job skeleton to customize, with the option to save it
to a JCL library, and run it.

The batch jobs are grouped into four major divisions by function as shown in
Figure 27 on page 127. All of the jobs, complete with sample JCL, are stored in
the SYS1.SACBCNTL library. Once you have selected the specific type of job to
be run, you can edit a copy of the sample JCL from an ISPF Edit panel.

Let′s look at an example using NaviQuest to set up a batch job, in this case,
creating a data set listing. On the Batch Testing/Configuration Selection Menu,
select ISMF option 1, Saved ISMF List Operations Batch Samples, to get to the
Saved ISMF List Operations Batch Samples Selection Menu (Figure 28 on
page 127).

126 DFSMS/MVS V1R4 Technical Guide


Panel Help

--------------------------------------------------------------------------------
ACBSMDJ1 BATCH TESTING/CONFIGURATION SELECTION MENU
Enter Selection or Command ===> 1______________________________________________

Select an option or enter Data Set to Edit and press Enter:

1 Saved ISMF List Operations Batch Samples

2 DCOLLECT Data Operations Batch Samples
3 Configuration Changes Batch Samples
4 VMA and SMF Batch Samples

Data Set to Edit . . ________________________________________________________

Use HELP Command for Help; Use END Command to Exit.

 
Figure 27. Batch Testing/Configuration Management Selection M e n u

The panel shows the names of the different sample jobs related to data set and
volume lists. You can scroll down to show three further jobs related to
producing tape volume lists. In our example, we type S next to Create Data Set
List.

Panel Help

--------------------------------------------------------------------------------
ACBSMDJ2 SAVED ISMF LIST OPERATIONS BATCH SAMPLES SELECTION MENU
Command ===> __________________________________________________________________

Select an option by typing ′ S′ or enter Data Set to Edit and press Enter:

More: +
S Create Data Set List
_ Create Data Set List and Save Query
_ Create Data Set List from a Saved Query
_ Generate Test Cases from a Data Set List
_ Generate Model Commands from a Saved List
_ Generate Data Set Report
_ Create Data Set List and Generate Data Set Report
_ Create DASD Volume List
_ Create DASD Volume List and Save Query
_ Create DASD Volume Query from a Saved Query
_ Generate DASD Volume Report
_ Create DASD Volume List and Generate DASD Volume Report

Data Set to Edit . . ________________________________________________________

Use HELP Command for Help; Use END Command to Exit.

 
Figure 28. Saved ISMF List Operations Batch Samples Selection M e n u

Pressing Enter will put you into ISPF Edit for the sample JCL. Once you have
modified the JCL and/or commands, you can submit the job. If you press PF3 (or

Chapter 9. NaviQuest and Catalog Search Interface 127

 END), a panel will be displayed asking you whether you want to save the
modified JCL. If you provide the name of the data set and member, you can
submit the same job in the future without having to go through the NaviQuest
panels.

If you want to define or modify the SMS configuration in batch, select option 3,
Configuration Changes Batch Samples, from the Batch Testing/Configuration
Selection Menu to get to the Configuration Changes Batch Samples Selection
Menu (Figure 29). This panel lists the sample jobs that you can select to define
or change various DFSMS policies in the DFSMS configuration. In our example,
we select Define/Alter Management Class.

Panel Help

--------------------------------------------------------------------------------
ACBSMDJ5 CONFIGURATION CHANGES BATCH SAMPLES SELECTION MENU
Command ===> __________________________________________________________________

Select an option by typing ′ S′ or enter Data Set to Edit and press Enter:

S Define /Alter Management Class

_ Define/Alter Pool Type Storage Group
_ Define/Alter Tape type Storage Group
_ Change Storage Group Volume Status
_ Translate ACS Routines
_ Validate SCDS
_ Test ACS Routines
_ Generate Enhanced ACS Test Listing
_ Compare ACS Test Listing
_ Update Test Cases with Test Results

Data Set to Edit . . ________________________________________________________

Use HELP Command for Help; Use END Command to Exit.

 
Figure 29. Configuration Changes Batch Samples Selection M e n u

Selecting this option will put you into ISPF Edit, and the JCL will be displayed as
shown in Figure 30 on page 129 and Figure 31 on page 130. Changes will have
to be made to the JOBCARD, prefix, and available parameters.
Here are the parameters for defining or altering a management class:
NaviQuest Key Word ISMF Field Name
SCDS CDS Name
MGMTCLAS Management Class Name
DESCR Description
EXPNOUSE Expire after Days Non-usage
EXPDTDY Expire after Date/Days
RETNLIM Retention Limit
PARTREL Partial Release
PRINOUSE Primary Days Non-usage
LV1NOUSE Level 1 Days Non-usage
CMDORAUT Command or Auto Migrate
PRIGDGEL # GDG Elements on Primary
GDGROLL Rolled-off GDS Action
BACKUPFR Backup Frequency
NUMBKDSE Number of Backup Vers
(Data Set Exists)

128 DFSMS/MVS V1R4 Technical Guide

 NUMBKDSD Number of Backup Vers
(Data Set Deleted)
RETDYDSD Retain days only Backup Ver
(Data Set Deleted)
RETDYEXT Retain days extra Backup Vers
CMDBKUP Admin or User command Backup
AUTOBKUP Auto Backup
BKUPTECH Backup Copy Technique

//IBMUSERA JOB (ACCT),′ IBMUSER′ , MSGCLASS=H,

// NOTIFY=IBMUSER,CLASS=A,MSGLEVEL=(1,1),TIME=(0,10)
//MYLIB JCLLIB ORDER=SYS1.SACBCNTL
//********************************************************************/
//*$MAC(ACBJBAJ1) COMP(5695DF123): BATCH - MC DEFINE/ALTER */
//* */
//* PROPRIETARY V3 STATEMENT */
//* LICENSED MATERIALS - PROPERTY OF IBM */
//* 5695-DF1 */
//* (C) COPYRIGHT 1996 IBM CORP. */
//* END PROPRIETARY V3 STATEMENT */
//* */
//* CHANGE ACTIVITY: */
//* */
//*$L0=NAVIQUEST,HACS120,96/06/18,SNJTCS: INITIAL VERSION @L0A*/
//* */
//********************************************************************/
//********************************************************************
//* *
//* SAMPLE JCL TO DEFINE/ALTER MANAGEMENT CLASSES IN BATCH *
//* *
//* INSTRUCTIONS BEFORE SUBMITTING: *
//* *
//* CHANGE JOBCARD *
//* CHANGE PREFIX *
//* CHANGE PARAMETERS *
//* *
//* PARAMETER FOLLOWING ACBQBAJ1 - DEFINE OR ALTER *
//* SCDS - SCDS IN WHICH MANAGEMENT CLASS IS TO BE DEFINED/ALTERED *
//* MGMTCLAS - MANAGEMENT CLASS TO BE DEFINED OR ALTERED *
//* *
//********************************************************************
//* *
//* STEP1 - SET UP THE PARAMETERS
//*
//********************************************************************
//STEP1 EXEC ACBJBAOB
//SYSUDUMP DD SYSOUT=*
//TEMPFILE DD DSN=&&TEMPFILE,DISP=(NEW,PASS),
// SPACE=(TRK,(1,1)),LRECL=300,RECFM=F,BLKSIZE=300

Figure 30. First Edit Screen

Chapter 9. NaviQuest and Catalog Search Interface 129

 //SYSTSIN DD *
PROFILE PREFIX(IBMUSER)
ISPSTART CMD(ACBQBAJ1 DEFINE +
SCDS(TEST.CDS) +
MGMTCLAS(STANDARD) +
DESCR(DEFINING STANDARD MGMT CLASS) +
EXPNOUSE() +
EXPDTDY() +
RETNLIM() +
PARTREL() +
PRINOUSE() +
LV1NOUSE() +
CMDORAUT() +
PRIGDGEL() +
GDGROLL() +
BACKUPFR() +
NUMBKDSE() +
NUMBKDSD() +
RETDYDSD() +
RETDYEXT() +
CMDBKUP() +
AUTOBKUP() +
BKUPTECH() +
)
//********************************************************************
//*
//* STEP2 - EXECUTE THE DEFINE/ALTER
//*
//********************************************************************
//STEP2 EXEC ACBJBAOB
//SYSUDUMP DD SYSOUT=*
//SYSTSIN DD DSN=&&TEMPFILE,DISP=(OLD,DELETE,DELETE)
//

Figure 31. Second Edit Screen. Shows the command and parameters that may be changed

For reference purposes the parameters for changing a pool storage group in
batch are:
NaviQuest Key Word ISMF Field Name
SCDS CDS Name
STORGRP Storage Group Name
DESCR Description
AUTOMIG Auto Migrate
MIGSYSNM Migrate Sys/Sys Group Name
AUTOBKUP Auto Backup
BKUPSYS Backup Sys/Sys Group Name
AUTODUMP Auto Dump
DMPSYSNM Dump Sys/Sys Group Name
DUMPCLAS Dump Class
HIGHTHRS Allocation/migration Threshold: High
LOWTHRS Allocation/migration Threshold: Low
GUARBKFR Guaranteed Backup Frequency
SGSTATUS SMS SG Status
DUMPCLAS can accept up to 5 values separated by commas.
SGSTATUS can accept up to 32 values separated by commas.

130 DFSMS/MVS V1R4 Technical Guide

 Note: You can use an asterisk (*) as a subparamter of SGSTATUS to leave the
status unmodified for a particular system or sysgroup. So, for example,
SGSTATUS(*,*,ENABLE) will change the status of the third system or sysgroup
alone, leaving unaltered the status of systems 1, 2, 4, and above.

For reference purposes the parameters for changing a tape storage group in
batch are:
NaviQuest Key Word ISMF Field Name
SCDS CDS Name
STORGRP Storage Group Name
DESCR Description
LIBNAME Library Names
SGSTATUS SMS SG Status
LIBNAME can accept up to 8 values separated by commas.
SGSTATUS can accept up to 32 values separated by commas.

Just as with a pool storage group, you can use an asterisk (*) as a subparameter
of SGSTATUS to leave the status unmodified for a particular system or sysgroup.

9.1.2 NaviQuest Help Panels

Field level help is provided for all NaviQuest panels. If you press PF1 on an
input field, help specific to that field will display. If the cursor is not on an input
field, panel help will display.

9.1.3 Migration and Coexistence

For those who have licensed IBM NaviQuest for MVS, the DFSMSdfp NaviQuest
component can use the test cases created. However, to use the new fields
introduced in DFSMS/MVS V1R3 in the ACS routines and to test for them, new
test cases must be created.

9.2 Catalog Search Interface

The CSI was developed as an MVS read-only general-use programming interface
(5799-CSI) to enable fast extraction of data from catalogs. It provides an
alternative to the AMS LISTCAT command. With DFSMS/MVS V1R4 CSI has
been incorporated into DFSMSdpf free of charge to make it available for all
customer and vendor applications.

The CSI is used to quickly obtain information about entries in the ICF catalogs.
The ICF catalog entries are selected by using a generic filter key provided as
input. The generic filter key can be either a fully qualified name that returns one
entry or a partially qualified name (wild cards) that returns multiple entries on a
single invocation. Several sample programs are provided to facilitate fast and
easy use of the CSI programming interface (see 9.2.5, “CSI Sample Programs”
on page 137).

9.2.1 Overview
To obtain information about ICF catalog entries, you request field information for
each entry by specifying field names. Thus you do not have to know whether the
information is in the Basic Catalog Structure (BCS) or the VSAM Volume Data
Set (VVDS).

Chapter 9. NaviQuest and Catalog Search Interface 131

 A work area, whose address and size are provided on input, is used to return
the selected entries and the field information for those entries. If the work area
cannot accommodate all of the entries, as many entries as possible are returned
in one invocation, and an indicator is set to reflect that more entries exist.
Subsequent invocations can specify that the request is being resumed, and the
additional entries can be obtained. This can be repeated until all entries have
been returne d. The resume process allows the ferrying of large amounts of
information to the user program without affecting ICF catalog resources.

9.2.2 Invocation
CSI can be invoked in either 24-bit or 31-bit addressing mode. CSI is reentrant
and reusable. CSI can be invoked in any protection key and in either supervisor
or problem state.

The invocation can be done through user application programs written in

assembler language or an HLL such as REXX.
• In an assembler program, CSI module IGGCSI00 can be invoked by one of
the following methods:
− A CALL statement that resolves to a V-type address constant
− A LINK macro
− LOAD and CALL macros
• In an HLL, CSI can be invoked by instructions that resolve to one of the
above mentioned assembler language invocations.

9.2.3 The Parameter List

On invocation, general purpose register (GPR) 1 points to a parameter list, GPR
13 points to a standard register save area 72 bytes long, GPR 14 contains the
return address in the caller′s program, and GPR 15 contains the entry point of
CSI.

Figure 32 on page 133 shows the CSI paramter list.

• The first word in the parameter list contains an address that points to a
4-byte reason area. On return from CSI, the reason area contains a module
identification, reason code, and return code.
• The second word contains the address of the selection criteria data fields.
These fields supply information to CSI on invocation and contain information
for resume if there is more informati on to be returned than can fit in the
user provided work area.
• The third word contains the address of a user work area in which entry
information will be returned. The caller must place the size of the work area
in the first word of the work area. Although this area has no fixed size, it
must be between 1024 bytes and 1,048,575 bytes, inclusive. CSI returns the
minimum size necessary to contain one entry, its catalog and fields included.
A reason code and return code will be set if a larger work area is needed.

132 DFSMS/MVS V1R4 Technical Guide

 ┌─────────┐
│ GPR 1 │
│ Address │
└───┬─────┘
│
│
└───── ┌───────────┬───────────┬───────────┐
│ Address 1 │ Address 2 │ Address 3 │
└─────┬─────┴─────┬─────┴─────┬─────┘
│ │ │
┌──────┘ │ └───┐
│ │ │
└─ ┌────────────┐ └─ ┌─────────┐ └─ ┌─────────┐
│Reason Area │ │CSIFLITK │ │Work Area│
└────────────┘ │CSICATNM │ │ │
│CSIRESNM │ │ │
│CSIDTYPS │ │ │
│CSIOPTS │ │ │
│CSINUMEN │ │ │
│CSIFLDNM │ │ │
│ ″ │ │ │
│ ″ │ └─────────┘
│ ″ │
└─────────┘
Figure 32. Catalog Search Interface Invocation Structure (Parameter List)

9.2.3.1 Reason Area

On return from CSI, the 4-byte reason area contains a module identifier, reason
code, and return code. The pattern of the information in the reason area is:
Description Length in Bytes
Module identification 2
Reason code 1
Return code 1

The contents of the reason area depend on the contents of GPR 15.

Refer to the “Return Status Information” section in the Catalog Search Interface
User ′ s Guide for a full explanation of return codes in GPR 15.

9.2.3.2 Selection Criteria Data Fields

The selection criteria data fields are your means of communicating with CSI.
These fields supply information to CSI during invocation and contain information
for a resume if there is more information to be returned than can fit in the
user-provided work area. The name of the selection criteria data fields are
described below.

Generic Filter Key Field (CSIFILTK): CSI uses a generic filter key supplied in
CSIFILTK. A generic filter key is a character string that describes the catalog
entry names for which you want information returned. The generic filter key can
contain the following symbols:

Chapter 9. NaviQuest and Catalog Search Interface 133

 * A single asterisk by itself indicates that either a qualifier or one or more
characters within a qualifier can occupy that position. An asterisk can
precede or follow a set of characters.
** A double asterisk indicates that zero or more qualifiers can occupy that
position. A double asterisk cannot precede or follow any character; it
must be preceded or followed by either a period or a blank.
% A single percent sign indicates that exactly one alphanumeric or national
character can occupy that position.
% % . . . One to eight percent signs can be specified in each qualifier.

Catalog Name (CSICATNM): CSICATNM is used for catalog selection. CSI uses
the catalog name supplied in the CSICATNM field to search for entries if
CSICATNM is not blank. If this field is blank, catalog management attempts to
use the high-level qualifier of CSIFILTK to locate an alias or multiple aliases that
match. If an alias is found, the user catalog for that alias is searched.
Otherwise, the master catalog is searched.

If a tape volume catalog library entry type or a tape volume catalog volume entry
type is specified in CSIDTYPS, a tape volume catalog is searched. Tape volume
catalog library entry types and tape volume catalog volume entry types should
not be mixed with ICF catalog entry types.

Resume Name (CSIRESNM): CSIRESNM is the name of the catalog entry on

which the search resumes if the work area space for return information is used
up. If the request can be resumed, CSIRESNM is set to Y on return from a call to
CSI and contains the name of the entry on which to resume, and CSICATNM
contains the name of the catalog in which the entry to be resumed was found.
Typically, the application program tests CSIRESNM; if it is Y, the application
program reissues the call to CSI without changing CSIRESNM and CSICATNM.

ICF Catalog Entry Types (CSIDTYPS): CSIDTYPS determines which type of

catalog entries will be returned. Valid types for CSIDTYPS are:
A Non-VSAM data set
B GDG
C Cluster
G Alternate index
H GDS
R Path
X Alias
U User catalog connector entry
L ATL library entry
W ATL volume entry

VSAM components data and index are returned with the cluster. Thus there are
no type specifications for them. However, D and I types will appear in the output
information.

The valid types can be mixed and in any order. Blanks cannot separate the
types. For instance, ABH might be specified to get only non-VSAM, GDG, and

134 DFSMS/MVS V1R4 Technical Guide

 GDS entries. All other positions in CSIDTYPS must be blank (X′40′) following the
specification of the entry types.

All blanks for CSIDTYPS can be set and will get types A, B, C, G, H, R, X, and U.
These are the ICF catalog types. L and W must be explicitly specified to get the
tape volume catalog entry.

CSI Options (CSIOPTS)

• CSICLDI determines whether information will be returned if the data and
index names of a cluster do not match the filter key but the cluster name
does. If CSICLDI is set to Y, the components are returned if the cluster name
matches the filter key.
• CSIRESUM is set to Y if CSI detects that there are more catalog entries than
meet the search criteria. This field must be blank on the first invocation.
CSIRESUM is tested, and CSI can be called again to obtain more entries.
CSIRESUM will be set to blank (X′40′) if no more entries meet the search
criteria.
• CSIS1CAT causes only one catalog to be searched. It is used in conjunction
with CSICATNM to determine which catalogs to search. This option is useful
if the filter key could cause catalog management to select more than one
catalog for searching. Also, if not set, the master catalog is searched in
addition to a selected user catalog. In cases where the catalog name is
supplied as input and resume is done, CSIS1CAT causes only that catalog
whose name is supplied as input to be searched. Otherwise, on resume,
catalog management cannot tell whether this is a search across many
catalogs and the resume caused the CSICATNM to be set or the name was
supplied by the caller.

Number of Field Names (CSINUMEN): CSINUMEN is a binary number indicating

the number of field names following in the subscripted area, CSIFLDNM.

It is possible to get a list of names without any field information. In this case, set
CSINUMEN to zero.

There is a limit of 100 field names per invocation of CSI. CSINUMEN cannot be
greater than 100.

Field Names (CSIFLDNM): CSIFLDNM is a list of 8-byte field names. If the field
name is not eight characters long, it must be padded on the right with blanks to
make eight characters. Refer to the “Field Name Directory” section in the
Catalog Search Interface User ′ s Guide for valid field names that can be used in
the list and the information returned for each field name.

9.2.4 Work Area

The third word contains the address of the work area in which entry information
will be returned. The caller must place the size of the work area in the first
word or the work area. Resume information is returned in the selection criteria
data fields. If on return CSIRESUM is set to Y, it will be set to the next entry to
be processed, and CSICATNM will be the catalog in which that entry will be
found. When the application program has processed returned entry information,
it should reinvoke CSI to resume at the next entry.

Chapter 9. NaviQuest and Catalog Search Interface 135

 The caller must set CSIUSRLN (length of user-provided work area) before
invoking CSI. CSIUSRLN must be a fixed fullword value between 1,024 and
1,048,575 bytes, inclusive.

9.2.4.1 Format Description

The first field is the length of the work area and is set by the user to tell CSI how
much data can be returned (see Figure 33 on page 137).

The second field is the minimum required length for a catalog entry and one
entry′s worth of returned data. If the minimum length is greater than the work
area length, then the work area length must be increased to at least as much as
the minimum length. If this is not done, and a resume condition occurs, the user
program will appear as if in an endless loop because the same information will
be returned for each resume until the first length is increased to contain the
entire entry. CSI sees a GDG with all of its associated GDSs as one record.
Thus the required length for this entry is apt to be large.

The third field is the amount of space that was used in the work area. CSI
always returns a full entry. If the last entry does not fully fit in the remaining
work area space, the resume flag is set and the space at the end of the work
area is unused. The unused space is usually small.

Next follows the number of field names plus one.

A catalog name entry is returned for every catalog processed. A catalog entry
can be identified because its type is X′F0′. This is an artificial type invented so
that the next catalog entry can be found. The catalog entry is always followed by
return information. The return code portion will be zero if problems were not
encountered while processing the catalog during the call.

Following the catalog entry is one or more entries contained in the catalog that
match the search criteria (filter key). Each entry has flags, followed by its type
and name. If the flags indicate, a module ID, reason code, and return code
follow the entry name; otherwise, the field information for the entry follows.

The MODULE ID / RSN / RC returned in the work area for each entry is
information returned by catalog management.

If no errors or messages occurred, the field information for the entry is returned
as a set of lengths with the data corresponding to the lengths.

The first length field is the total length of the length fields including this field and
all of the data returned for this entry. The total length field is 2 bytes long. After
that is a reserved 2-byte field.

Next are a set of lengths corresponding to the number of fields passed in. Each
length is used to determine the length and position of the returned data of the
entry. For example, if three field names were supplied on input, there will be
three field lengths. Each length will be for the data immediately following the
lengths. If the lengths had values 4, 6, and 8, then, following the last length,
there would be 4 bytes worth of data for the first field, 4 bytes from the last
length field would be 6 bytes of data for the second field, and 10 bytes from the
last length would be 8 bytes worth of data for the third field.

136 DFSMS/MVS V1R4 Technical Guide

 WORK AREA
┌────────────────────────────────┐
│ USER PROVIDED WORK AREA LENGTH │
├────────────────────────────────┤
│ REQUIRED WORK AREA LENGTH │
├────────────────────────────────┤
│ USED USER WORK AREA LENGTH │
├────────────────────────────────┤
│ NUMBER OF FIELD NAMES + 1 │
├──────┬──────┬──────────────────┤
│ FLAG │ TYPE │ CATALOG NAME │
├──────┴──────┼───────┬──────────┤
│ MODULE ID │ RSN │ RC │
├──────┬──────┼───────┴──────────┤
│ FLAG │ TYPE │ ENTRY NAME │
├──────┴──────┼───────┬──────────┤
│ MODULE ID │ RSN │ RC │
├─────────┬───┴────┬──┴──────────┤
│ TOT LEN │RESERVED│ LENGTH FLDS │
├─────────┴────────┴─────────────┤
│ │
│ │
/ FIELD DATA /
│
├────────────────────────────────┐
│ │
│ │
Figure 33. The Catalog Search Interface Work Area

9.2.5 CSI Sample Programs

Three sample assembler programs and one sample REXX EXEC come with the
CSI product. You can use the sample programs to try out CSI without having to
write your own program. You should be able to get CSI up and running, using
these programs with only a few JCL changes. The sample programs are
delivered as is and are not intended to be part of the licensed product. You may
want to modify them to fit your specific needs.

IGGCSILC: IGGISLC is an assembler program that produces output similar to

AMS (IDCAMS) LISTCAT NAME.

IGGCSIVG: IGGCSIVG is an assembler program that identifies unused space at

the end of a VSAM data set. Basically, it computes the difference of the
high-used and the high-allocated RBAs for each VSAM data set in a given
catalog. This program is useful for identifying overallocated space.

IGGCSIVS: IGGCSIVS is an assembler program that identifies which data sets

on a given volume reside in a particular catalog. In the event of a disk drive
failure, this program would be useful in identifying which entries in a catalog
have to be cleaned up if the data sets were recovered to a different volser.

IGGCSIRX: IGGCSIRX is a REXX EXEC that uses CSI. When executed,

IGGCSIRX prompts you for a filter key. This should be a partially qualified data
set name as described for the selection criteria data field CSIFILTK. The data
set name, its type, and volser are returned to your TSO session.

Chapter 9. NaviQuest and Catalog Search Interface 137

138 DFSMS/MVS V1R4 Technical Guide
Appendix A. Parallel Sysplex and VSAM Record Level Sharing

VSAM RLS extends the DFSMS/MVS storage hierarchy to support data sharing
across multiple systems in a System/390 parallel sysplex. Enhancements in
DFSMS/MVS V1R4 take advantage of VSAM RLS data access and thus exploit
the capabilities of a parallel sysplex. In this appendix we review VSAM RLS and
parallel sysplex technology.

A.1 Overview of Parallel Sysplex

A parallel sysplex is a collection of MVS/ESA systems that cooperate using
certain hardware and software products to process work in multiple CPCs.
Hardware required for a parallel sysplex includes:
• The Sysplex Timer for clock synchronization
• Shared DASD
• One or more coupling facilities and coupling facility links for enabling
parallel processing and improved data sharing

The benefits of a parallel sysplex include:

• High-performance data sharing with integrity
• Single system image and point of control
• Continuous availability
• Reduced cost of computing
Note: The coupling facility effectively becomes another part of the storage
hierarchy. It is shared by all processors in the sysplex and enables data buffers
and locks to be shared across the sysplex.

Major transaction and database systems that can exploit the coupling facility
technology within a parallel sysplex are IMS/ESA, DB2, and CICS VSAM RLS.
DFSMShsm can now also exploit the functions and performance of VSAM RLS by
allowing the migration, backup, and offline control data sets (MCDS, BCDS, and
OCDS) to be cached in the coupling facility.

A.2 Overview of VSAM Record Level Sharing

VSAM RLS, a data set accessing mode introduced in DFSMS/MVS V1R3, exploits
the locking and caching capabilities of the coupling facility in a parallel sysplex
environment. It enables VSAM data to be shared with full update capability
among many applications running in many CICS regions.

Other subsystems, such as DFSMShsm, can exploit the performance and

functions of VSAM RLS.

VSAM RLS relies on the following four functions, which are provided by other
software:
• Central logging
VSAM data can be accessed by multiple users. For integrity, backup, and
recovery purposes, the exploiting subsystem software can choose to log
changes to the VSAM data set by using the MVS Logger.
• Central locking

 Copyright IBM Corp. 1997 139

 Locking must be provided from a central facility, which is used by all
subsystems accessing VSAM data.
• Logical unit of work control
Unit of work control must be understood by all parties so that they do not
work with uncommitted data.
• Retained locks
Active locks must be retained if a subsystem fails and leaves updated
records in an uncommitted state. Retained locks ensure that data integrity is
maintained until the failing subsystem (for example, CICS, HSM) has
completed its recovery processing.

To use VSAM RLS, the VSAM data set must be SMS managed. The storage
class assigned to the VSAM data set determines which cache structures to use
in the coupling facility, if the VSAM data set is opened in RLS mode.

VSAM RLS is an access mode that determines a particular manner of buffer

management and control, and is mutually exclusive with NSR, LSR, or GSR
buffer management. The RLS option is specified by either using a new JCL
parameter (RLS= ) on the DD card or specifying MACRF=RLS in the ACB.

VSAM RLS supports the KSDS, ESDS, RRDS, and VRRDS file organizations.
KSDS and ESDS access is also supported through path access.

RLS specifies that VSAM record level sharing protocols are used and implies
that VSAM uses cross-system record level locking, as opposed to CI locking,
uses the coupling facility for buffer consistency and performance across the
sysplex, and manages a systemwide local cache. With VSAM RLS the coupling
facility is used as a store through cache and so the data on the shared DASD
always reflects the most recent updates to the data set. When a VSAM data set
is accessed in RLS mode, the SHAREOPTION values are totally ignored as the
VSAM RLS code always assumes multiple readers and writers and takes
responsibility for data integrity.

Currently, when a VSAM data set is opened in RLS mode, it cannot be accessed
by any other application in any other mode (NSR, LSR, or GSR) at the same
time. NSR reads will be allowed by 3Q97.

DFSMS supports a data sharing subsystem, first introduced in DFSMS/MVS

V1R3, called SMSVSAM, which runs in its own address space to provide the RLS
support required by the VSAM RLS applications within each MVS image in a
parallel sysplex environment.

The SMSVSAM server uses the coupling facility for its cache and lock structures.
With VSAM RLS, the application buffers and most VSAM control blocks are
allocated in the associated 2GB data space with the SMSVSAM address space.
The only VSAM control blocks still in the application address space are the ACB,
RPLs, and EXLST.

The SMSVSAM servers (one per MVS image) coordinate with each other to
logically maintain a single control block structure for each open VSAM RLS data
set across the sysplex and assumes responsibility for maintaining
synchronization across the sysplex.

140 DFSMS/MVS V1R4 Technical Guide

A.3 VSAM RLS Locking
VSAM RLS allows VSAM data to be shared with locking performed at the record
level across the sysplex. Thus multiple transactions can update different records
even within the same CI without waiting for one another.

Because a data set CI can reside in more than one local buffer, it can be
updated concurrently. Because record-level locking is performed, any
concurrent updates are against different records in the same CI. The changed
records from the multiple copies of the CI are merged to create a copy of the CI
that contains all of the changed records. This avoids the update problem where
an update can be lost when multiple independent tasks attempt to update the
same block of data from their own buffers, not realizing that they do not have
exclusive control over this block of data.

Locking at the record level rather than at the CI, CA, or data set level provides a
much finer granularity of locking than would otherwise be the case. This helps
performance by limiting exclusive lockouts to cases where multiple applications
attempt to update the same record.

Appendix A. Parallel Sysplex and VSAM Record Level Sharing 141

142 DFSMS/MVS V1R4 Technical Guide
Appendix B. Special Notices

This publication is intended to provide storage administrators, system

programmers, and experienced storage personnel with a technical preview of
the functions and enhancements of DFSMS/MVS V1R4. The information in this
publication is not intended as the specification of any programming interfaces
that are provided by DFSMS/MVS V1R4. See the PUBLICATIONS section of the
IBM Programming Announcement for DFSMS/MVS V1R4 for more information
about what publications are considered to be product documentation.

References in this publication to IBM products, programs or services do not

imply that IBM intends to make these available in all countries in which IBM
operates. Any reference to an IBM product, program, or service is not intended
to state or imply that only IBM′s product, program, or service may be used. Any
functionally equivalent program that does not infringe any of IBM′s intellectual
property rights may be used instead of the IBM product, program or service.

Information in this book was developed in conjunction with use of the equipment
specified, and is limited in application to those specific hardware and software
products and levels.

IBM may have patents or pending patent applications covering subject matter in
this document. The furnishing of this document does not give you any license to
these patents. You can send license inquiries, in writing, to the IBM Director of
Licensing, IBM Corporation, 500 Columbus Avenue, Thornwood, NY 10594 USA.

Licensees of this program who wish to have information about it for the purpose
of enabling: (i) the exchange of information between independently created
programs and other programs (including this one) and (ii) the mutual use of the
information which has been exchanged, should contact IBM Corporation, Dept.
600A, Mail Drop 1329, Somers, NY 10589 USA.

Such information may be available, subject to appropriate terms and conditions,

including in some cases, payment of a fee.

The information contained in this document has not been submitted to any
formal IBM test and is distributed AS IS. The use of this information or the
implementation of any of these techniques is a customer responsibility and
depends on the customer′s ability to evaluate and integrate them into the
customer′s operational environment. While each item may have been reviewed
by IBM for accuracy in a specific situation, there is no guarantee that the same
or similar results will be obtained elsewhere. Customers attempting to adapt
these techniques to their own environments do so at their own risk.

Reference to PTF numbers that have not been released through the normal
distribution process does not imply general availability. The purpose of
including these reference numbers is to alert IBM customers to specific
information relative to the implementation of the PTF when it becomes available
to each customer according to the normal IBM PTF distribution process.

The following terms are trademarks of the International Business Machines

Corporation in the United States and/or other countries:

AIX BookManager
BookMaster C/370

 Copyright IBM Corp. 1997 143

 CBIPO CICS
Common User Access CUA
DB2 DFSMS/MVS
DFSMSdfp DFSMSdss
DFSMShsm DFSMSrmm
Hiperspace IMS
MVS/ESA OpenEdition
OS/2 OS/390
Parallel Sysplex RACF
RAMAC Resource Measurement Facility
RMF RS/6000
S/390 Sysplex Timer
System/390

The following terms are trademarks of other companies:

C-bus is a trademark of Corollary, Inc.

PC Direct is a trademark of Ziff Communications Company and is

used by IBM Corporation under license.

UNIX is a registered trademark in the United States and other

countries licensed exclusively through X/Open Company Limited.

Microsoft, Windows, and the Windows 95 logo

are trademarks or registered trademarks of Microsoft Corporation.

Java and HotJava are trademarks of Sun Microsystems, Inc.

Other trademarks are trademarks of their respective companies.

144 DFSMS/MVS V1R4 Technical Guide

Appendix C. Related Publications

The publications listed in this section are considered particularly suitable for a
more detailed discussion of the topics covered in this redbook.

C.1 International Technical Support Organization Publications

For information on ordering these ITSO publications see “How to Get ITSO
Redbooks” on page 147.
• DFSMS Optimizer Presentation Guide , SG24-4477
• Converting to RMM: A Practical Guide , SG24-4998
• DFSMS/MVS V1R3.0 Presentation Guide , GG24-4391
• NaviQuest Demonstration and Hands-On Usage Guide , SG24-4720

C.2 Redbooks on CD-ROMs

Redbooks are also available on CD-ROMs. Order a subscription and receive
updates 2-4 times a year at significant savings.

CD-ROM Title Subscription Collection Kit

Number Number
System/390 Redbooks Collection SBOF-7201 SK2T-2177
Networking and Systems Management Redbooks Collection SBOF-7370 SK2T-6022
Transaction Processing and Data Management Redbook SBOF-7240 SK2T-8038
AS/400 Redbooks Collection SBOF-7270 SK2T-2849
RS/6000 Redbooks Collection (HTML, BkMgr) SBOF-7230 SK2T-8040
RS/6000 Redbooks Collection (PostScript) SBOF-7205 SK2T-8041
Application Development Redbooks Collection SBOF-7290 SK2T-8037
Personal Systems Redbooks Collection SBOF-7250 SK2T-8042

C.3 Other Publications

These publications are also relevant as further information sources:
• Remote Copy Administrator ′ s Guide and Reference , SC35-0169
• DFSMS/MVS V1R3 NaviQuest User ′ s Guide , SC26-7194
• DFSMS Optimizer User ′ s Guide and Reference , SC26-7047
• Using the Interactive Storage Management Facility , SC26-4911
• DFSMS/MVS V1R3 DFM/MVS Guide and Reference , SC26-4915
• DFSMS/MVS V1R3 NFS User ′ s Guide , SC26-7028
• DFSMS/MVS V1R3 NFS Customization and Operation , SC26-7029
• Catalog Search Interface User ′ s Guide , SC26-7015
• DFSMS/MVS V1R3 Access Method Services for ICF , SC26-4906
• DFSMS/MVS V1R3 Installation Exits , SC26-4908
• DFSMS/MVS V1R3 DFSMSdfp Storage Administration Reference , SC26-4920

 Copyright IBM Corp. 1997 145

146 DFSMS/MVS V1R4 Technical Guide
How to Get ITSO Redbooks
This section explains how both customers and IBM employees can find out about ITSO redbooks, CD-ROMs,
workshops, and residencies. A form for ordering books and CD-ROMs is also provided.

This information was current at the time of publication, but is continually subject to change. The latest
information may be found at http://www.redbooks.ibm.com.

How IBM Employees Can Get ITSO Redbooks

Employees may request ITSO deliverables (redbooks, BookManager BOOKs, and CD-ROMs) and information about
redbooks, workshops, and residencies in the following ways:
• PUBORDER — to order hardcopies in United States
• GOPHER link to the Internet - type GOPHER.WTSCPOK.ITSO.IBM.COM
• Tools disks
To get LIST3820s of redbooks, type one of the following commands:
TOOLS SENDTO EHONE4 TOOLS2 REDPRINT GET SG24xxxx PACKAGE
TOOLS SENDTO CANVM2 TOOLS REDPRINT GET SG24xxxx PACKAGE (Canadian users only)
To get BookManager BOOKs of redbooks, type the following command:
TOOLCAT REDBOOKS
To get lists of redbooks, type one of the following commands:
TOOLS SENDTO USDIST MKTTOOLS MKTTOOLS GET ITSOCAT TXT
TOOLS SENDTO USDIST MKTTOOLS MKTTOOLS GET LISTSERV PACKAGE
To register for information on workshops, residencies, and redbooks, type the following command:
TOOLS SENDTO WTSCPOK TOOLS ZDISK GET ITSOREGI 1996
For a list of product area specialists in the ITSO: type the following command:
TOOLS SENDTO WTSCPOK TOOLS ZDISK GET ORGCARD PACKAGE
• Redbooks Web Site on the World Wide Web
http://w3.itso.ibm.com/redbooks
• IBM Direct Publications Catalog on the World Wide Web
http://www.elink.ibmlink.ibm.com/pbl/pbl
IBM employees may obtain LIST3820s of redbooks from this page.
• REDBOOKS category on INEWS
• Online — send orders to: USIB6FPL at IBMMAIL or DKIBMBSH at IBMMAIL
• Internet Listserver
With an Internet e-mail address, anyone can subscribe to an IBM Announcement Listserver. To initiate the
service, send an e-mail note to [email protected] with the keyword subscribe in the body of
the note (leave the subject line blank). A category form and detailed instructions will be sent to you.

Redpieces

For information so current it is still in the process of being written, look at ″Redpieces″ on the Redbooks Web
Site ( http://www.redbooks.ibm.com/redpieces.htm). Redpieces are redbooks in progress; not all redbooks
become redpieces, and sometimes just a few chapters will be published this way. The intent is to get the
information out much quicker than the formal publishing process allows.

 Copyright IBM Corp. 1997 147

How Customers Can Get ITSO Redbooks
Customers may request ITSO deliverables (redbooks, BookManager BOOKs, and CD-ROMs) and information about
redbooks, workshops, and residencies in the following ways:
• Online Orders — send orders to:

IBMMAIL Internet
In United States: usib6fpl at ibmmail [email protected]
In Canada: caibmbkz at ibmmail [email protected]
Outside North America: dkibmbsh at ibmmail [email protected]

• Telephone orders

United States (toll free) 1-800-879-2755

Canada (toll free) 1-800-IBM-4YOU

Outside North America (long distance charges apply)

(+45) 4810-1320 - Danish (+45) 4810-1020 - German
(+45) 4810-1420 - Dutch (+45) 4810-1620 - Italian
(+45) 4810-1540 - English (+45) 4810-1270 - Norwegian
(+45) 4810-1670 - Finnish (+45) 4810-1120 - Spanish
(+45) 4810-1220 - French (+45) 4810-1170 - Swedish

• Mail Orders — send orders to:

I B M Publications I B M Publications IBM Direct Services

Publications Customer Support 144-4th Avenue, S.W. Sortemosevej 21
P.O. Box 29570 Calgary, Alberta T2P 3N5 DK-3450 Allerød
Raleigh, NC 27626-0570 Canada Denmark
USA

• Fax — send orders to:

United States (toll free) 1-800-445-9269

Canada 1-403-267-4455
Outside North America (+45) 48 14 2207 (long distance charge)

• 1-800-IBM-4FAX (United States) or (+1)001-408-256-5422 (Outside USA) — ask for:

Index # 4421 Abstracts of new redbooks
Index # 4422 IBM redbooks
Index # 4420 Redbooks for last six months
• Direct Services - send note to [email protected]
• On the World Wide Web
Redbooks Web Site http://www.redbooks.ibm.com
IBM Direct Publications Catalog http://www.elink.ibmlink.ibm.com/pbl/pbl
• Internet Listserver
With an Internet e-mail address, anyone can subscribe to an IBM Announcement Listserver. To initiate the
service, send an e-mail note to [email protected] with the keyword subscribe in the body of
the note (leave the subject line blank).

Redpieces

For information so current it is still in the process of being written, look at ″Redpieces″ on the Redbooks Web
Site ( http://www.redbooks.ibm.com/redpieces.htm). Redpieces are redbooks in progress; not all redbooks
become redpieces, and sometimes just a few chapters will be published this way. The intent is to get the
information out much quicker than the formal publishing process allows.

148 DFSMS/MVS V1R4 Technical Guide

IBM Redbook Order Form
Please send me the following:

Title Order Number Quantity

First name Last name

Company

Address

City Postal code Country

Telephone number Telefax number VAT number

• Invoice to customer number

• Credit card number

Credit card expiration date Card issued to Signature

We accept American Express, Diners, Eurocard, Master Card, and Visa. Payment by credit card not
available in all countries. Signature mandatory for credit card payment.

How to Get ITSO Redbooks 149

150 DFSMS/MVS V1R4 Technical Guide
List of Abbreviations
ABARS aggregate backup and recovery support ISMF Interactive Storage Management Facility
ABR aggregate backup and recovery ISV independent software vendor
ACB access control block ITSO International Technical Support Organization
AMP access method parameter JFCB job file control block
APA all points addressable KSDS key-sequenced data set
API application programming interface LACS layout analysis and checking system
APPC Advanced Program-to-Program LAN local area network
Communication
LCS library control system
ATL automated tape library
LFS logical file system
BCDS backup control data set
LLA library lookaside
BCS basic catalog structure
LPA link pack area
BWO backup while open
LPAR logically partitioned mode
CA control area
LRD last reference date
CCW channel control word
LSR local shared resource
CDS control data set
MCDS migration control data set
CI control interval
MWE management work element
CPC central processing complex
NFS Network File System
CSI catalog search interface
NSR nonshared resource
DADSM direct access device space management
OAM Object Access Method
DAE dump analysis elimination
OCDS output control data set
DAS data access services
OCO object code only
DCB data control block
OSMC OAM Storage Management Component
DFDSS Data Facility Data Set Services
OSR object storage and retrieval
DFHSM Data Facility Hierarchical Storage Manager
PDA problem determination aid
DFM Distributed FileManager
PDS partitioned data set
DSCB data set control block
PFS physical file system
EA extended addressibility
PDSE partitioned data set extended
EF extended format
PO partitioned organization (value of DSORG)
FCT file control table
PPRC peer-to-peer remote copy
GDG generation data group
PROFS Professional Office System
GDS generation data set
PS physical sequential (value of DSORG)
GPR general purpose register
PSU physical sequential unmovable (value of
GRS global resource serialization DSORG)
GTF generalized trace facility RBA relative byte address
FEOV force end of volume RLD relocation dictionary
HLL high level language RLS record level sharing
HLQ high level qualifier SAF security authorization facility
IBM International Business Machines SDM system data mover
Corporation
SDR sustained data rate
ICF Integrated Catalog Facility
SdU SMARTdata UTILITIES
IOST I/O summary trace
SDWA system diagnostic work area

 Copyright IBM Corp. 1997 151

SMB system-managed buffering UTC universal time, coordinated
SMS system-managed storage VRS vital record specification
STCK store clock VRSEL vital record selection
SVC supervisor call instruction VVDS VSAM volume data set
TCN transportation control number WTOR write to operator with reply
TIOT task input/output table WWFSR ABACKUP/ARECOVER function statistics
record
TMM tape mount management
XCF cross system coupling facility
TTOC tape table of contents
XRC extended remote copy
UCB unit control block

152 DFSMS/MVS V1R4 Technical Guide

Index
catalog search interface (continued)
A CSIFLDNM option 135
ABARS CSINUMEN option 135
ABR record 60 CSIOPTS 135
activity log 59 CSIRESNM filter key field 134
ALLOCATE statement GDG base name 59 enhancement overview 4
ARCBEEXT exit 58 generic filter field 133
concurrent ABARS requests 58 IGGCSILC program 137
CPU time in WWFSR 60 IGGCSIRX program 137
enhancements 57 IGGCSIVG program 137
ISMF changes 61 IGGCSIVS program 137
migration considerations 61 invocation 132
OPTIMIZE keyword externalized 61 o v e r v i e w 131
output file stacking 57 parameter list 132
SMF record for WWFSR 60 reason area 133
TGTGDS keyword externalized 60 resume name field 134
ABARS activity log sample programs 137
delete during roll-off processing 59 selection criteria fields 133
ABARS output file stacking work area 135
ABACKUP enhancement 58 work area format 136
ABARSTAPES parameter checkpointed data sets
ARECOVER enhancement 58 ADRUIXIT exit 27
invocation CONVERTV processing 24
NOSTACK keyword 58 COPY DATASET processing 25
STACK keyword 58 copy/dump processing 24
ABARSTAPES parameter 58 current access description 22
ABR record 60 DEFRAG processing 25
ACB MACRF parameters 36 DFSMShsm BACKUP/ABACKUP 28
ADRUIXIT exit 27, 76 DFSMShsm migration 28
alter without recall 55 DFSMShsm patch 27
catalog error processing 57 DFSMShsm recall 28
catalog notification 56 DFSMShsm RECOVER/ARECOVER 28
catalog processing considerations 56 DS1CPOIT indicator 22
HALTER processing 56 DS1DFGU indicator 23
IDCAMS calalog interface 56 DS1REFD indicator 23
o v e r v i e w 55 logical data set dump 25
ARCBEEXT exit 58 logical data set restore 26
ARCIMPRT CDS recovery function 65 migration considerations 28
ARCPRPDO utility 88 O/C/EOV enhancement 24
ARCTVEXT installation exit 67 physical data set dump 25
RELEASE processing 26
volume dump 28
B concurrent ABARS requests
BACKUPPROC procedure 80
64 concurrent requests 58
BPXPRMxx PARMLIB member 97
invocation 58
started task identifier 58
concurrent copy 83
C
catalog search interface
catalog entry types field 134
catalog name field 134
D
data class attributes
CSI options 135
BWO 44, 45
CSICATNM filter key field 134
LOG 44
CSIDTYPS filter key field 134
LOGSTREAM ID 44, 46
CSIFILTK filter key field 133
NONSPANNED 44, 47

 Copyright IBM Corp. 1997 153

data class attributes (continued) DFSMSdss
SPANNED 44, 46 ADRUIXIT exit 76
DataAgent routines 102 ALLMULTI keyword 73
DCOLLECT enhancements CONVERTV processing 75
DC record 48 data mover for DFSMSrmm CDS backup 83
DCACICA field 48 migration considerations 77
DCACISZ field 48 multivolume selection 73
DCOLLECT command 48 SELECTMULTI keyword 73, 74
DCVDPTYP field 48 DFSMShsm
DDCRABS bit 48 ABARS activity log 59
DDCRBIAS bit 48 ABARS enhancements 57
EXCLUDEVOLUMES parameter 47, 48 alter without recall 55
UMBFRVO field 48 alternative to TAPECOPY processing 51
UMFRVO field 48 ARCBEEXT exit 58
DDM protocol 101 ARCIMPRT CDS recovery function 65
DFM DataAgent ARCPRPDO utility 88
DDMClose function 107 AUDIT DATASETCONTROLS 71
DDMOpen function 106 CDS access in RLS mode 62
DFM target 108 checkpoint/restart patch 27
DFM00 PARMLIB member 112 concurrent ABARS requests 58
DFMACALL application 111 DAE utilization 69
DFMQTSO routine 111 DFSMShsm enhancement overview 51
DFMX0001 routine 111 DFSMSrmm interface to DFSMShsm 67
DFMXAGNT routine 112 duplex tape 51
DFMXLPRM system service 110 IDCAMS calalog interface 56
DFMXTSO routine 111 keyword aliases 71
file name suffix parameters 107 LOCATE macro 56
interfaces 108 Optimizer interface 72
invocation 105 output file stacking 57
o v e r v i e w 102, 105 PDA formatter 88
parameter list 109 SMF FSR subtypes 70
routine considerations 110 TAPECOPY processing changes 53
routines 102 TCN record 52
system messages 113 DFSMShsm CDS sharing 66
DFM00 PARMLIB member 112 ARCIMPRT CDS recovery function 65
DFSMS FIT 120 CDS backup in RLS access mode 65
DFSMS/MVS NFS coexistence considerations 66
client commands 99 DCOLLECT enhancements 65
client configuration 97 RLS mode access 62
client log data sets 98 RLS mode considerations 63
deleting migrated data sets 100 RLS mode invocation 62
GFSCPROC 97 RLS mode sample JCL 63
LFS colony address space 98 DFSMShsm DAE utilization
mounting remote file systems 98 ABARS secondary address space 70
NFS client function 96 description 69
obtaining migrated data set attributes 100 eligible dumps 69
OpenEdition access method support 99 implementation 69
OpenEdition interface 98 primary address space 70
physical file system 98 DFSMShsm serviceability items
server support 100 AUDIT DATASETCONTROLS 71
supported client functions 96 GFTEMINX interface 72
DFSMS/MVS NFS client commands 99 keyword aliases 71
DFSMSdfp Optimizer interface 72
checkpointed data sets 22 SMF FSR subtypes 70
enabling tailored compression 15 DFSMSrmm
O/C/EOV serviceability enhancements 19 DFSMShsm alternate tape support 88
Program Management 3 (PM3) 32 EDGHSKP parameters 93
space allocation failure reduction 7 EDGRMMxx PARMLIB member 79, 91

154 DFSMS/MVS V1R4 Technical Guide

DFSMSrmm (continued) DS1CPOIT indicator 22
EDGUX100 exit 89 DS1DSGU unmovable indicator 23
external data managers 89 DS1REFD 41
inventory management processing 91 duplex tape 51
journal usage threshold 79
o v e r v i e w 85
PDA trace 85 E
VRSMIN option 92 EDGHSKP 93
DFSMSrmm interface to DFSMShsm EDGPDOX log data set 85
ARCTVEXT installation exit 67 EDGPDOY log data set 85
EDGDFHSM program interface 67 EDGRMMxx PARMLIB member 91
EDGTVEXT program interface 67 EDGTVEXT program interface 67
interface processing 67 EDGUX100 exit 89
migration considerations 68 EDGUX100 exit sample 90
DFSMSrmm inventory management
ACTIVITY file 92
EDGHSKP parameters 93
F
FORCEP parameter 24
inventory management processing 91
messages 94
migration considerations 93
VRSEL processing 90
G
generic compression 13
VRSMIN option 92
GFSCPROC 97
DFSMSrmm journal usage threshold
automate backup procedure 81
automate journal clearing 80 I
BACKUPPROC procedure 80, 81 IEAAPFxx PARMLIB member 97
EDGRMMxx PARMLIB member 79 IFAPRDxx PARMLIB member 97
messages 81 IFG0554P module 19
threshold setting 80 IFGOCETR started task 20
TSO subcommand variables 82 IGDACSSC exit 122
DFSMSrmm nonintrusive CDS backup IGDDFPKG PARMLIB member 97
CDS backup 82 IGDSMSxx PARMLIB member 15
concurrent copy 83 IGG026DU 56
DFSMSdss as data mover 83 IGGCSILC program 137
DFSMSdss CDS backup considerations 84 IGGCSIRX program 137
DFSMSdss DUMP command 83 IGGCSIVG program 137
DFSMSdss parameter keywords 84 IGGCSIVS program 137
DFSMSdss RESTORE command 83 inventory management processing 91
DSSOPT DD statement 83 ISMF data class panels 37
hardware requirements 85
DFSMSrmm PDA trace
ARCPRPDO utility 88 L
DFSMSrmm startup procedure 87 last reference date at CLOSE 41
EDGPDOX log data set 85 LFS colony address space 98
in-storage circular file 86 LOCATE macro 56
migration considerations 87 LSR resource pools 39
PDA formatter 88
performance 88
trace data set requirements 87 M
trace data set sizing 86 messages
trace implementation 87 ADR211I 25
Distributed FileManager ADR297I 25
DataAgent routine considerations 110 ADR298E 25, 26
DDM protocol 101 ARC1245I 28
DFM DataAgent 102 EDG2103E 81
DFMXLPRM system service 110 EDG2104E 81
SMARTdata UTILITIES 103 EDG2107E 79
supported MVS data set types 101 EDG2108E 82
IEC026I 19

Index 155
messages (continued) OSREQ macro 31
IEC161I 43
IEC214I 19
IEC813I 20 P
IEC980A 20, 21 Parallel Sysplex
IEC980I 19 benefits 139
multivolume selection 73 definition 139
exploiters 139
hardware required 139
N PDA trace 85
NaviQuest physical file system 98
ACS management option panels 120 Program Management 3 (PM3)
ACS test listings comparison 124 background 32
define/alter management class in batch 129 Dynamic Load Libraries 33
DFSMS FIT 120 enhancement description 33
IGDACSSC exit 122 OpenEdition support 33
ISMF functions in batch 126
ISMF saved lists 120
migration considerations 131 R
o v e r v i e w 119 record access bias
regression testing 120 buffer optimization 35
SAVE TSOLIST command 120 buffering handling 36
SYS1.SACBCNTL library 126 data class specification 36
test case generation 121 ISMF data class panels 37
Network File System 95 JCL specification 36
NOSTACK keyword 58 record level sharing
DFSMShsm exploitation 62
RLS KSDS extended addressability
O data class definition 42
O/C/EOV serviceability enhancements hardware requirements 44
abend message 19 JCL specification 44
activating IFGOCETR 20 migration considerations 43
DFSMSdfp enhancements 23 software prerequisites 43
DFSMSdss enhancements 24 RMODE31 JCL AMP parameter 38
DFSMShsm enhancements 27
FORCEP parameter 24
IFGOCETR parameters 21 S
IFGOCETR started task 20 SAM tailored compression 12
migration considerations 22, 28 SETSYS command
OAM SMF recording enhancements 29 ABARSDELETEACTIVITY parameter 59
o v e r v i e w 19 ABARSOPTIMIZE keyword 61
PDS resource held for output 20 ABARSTAPES parameter 58
SMF recording 20 ARECOVERTGTGDS parameter 60
trace processing 20 CDSVERSIONBACKUP parameter 66
OAM SMF recording DUPLEX parameter 53
activating 31 EXITOFF parameter 67
OSREQ macro 31 EXITON parameter 58, 67
SMF record format 29 MAXABARSADDRESSSPACE parameter 58
SMF record subtypes 29 NOUSERUNITTABLE parameter 71
TTOKEN keyword 31 SMF parameter 60
OpenEdition access method support 99 SYS1DUMP parameter 69
OPTIMIZE keyword 61 USERUNITTABLE parameter 71
Optimizer SMARTdata UTILITIES 103
HSM monitor/tuner 117 addressing distributed data 104
input data 115 calling the utilities 104
management class analyzer 117 client/server development 104
o v e r v i e w 115 data access functions 105
performance analyzer 116 data description and conversion 103
DFM 103

156 DFSMS/MVS V1R4 Technical Guide

SMARTdata UTILITIES (continued) VSAM RLS 139
DFMACALL application 111 central locking 139
sharing distributed data 105 central logging 139
SMARTsort 103 locking 141
VSAM local file system 103 retained locks 140
space allocation failure reduction 9 SMSVSAM server 140
allocations affected 8 supported data set types 140
best-fit allocation 9 unit of work control 140
data class fields 8 usage requirements 140
description 7
extents on new volumes 10
extents per VSAM component 8 W
implementation 8 wellness logic, compression 15
ISMF changes 11
ISMF messages 11
migration and coexistence 11
new allocations 9
non-VSAM extent processing 9, 10
spreading data 8
VSAM extent processing 9, 10
STACK keyword 58
SYS1.SACBCNTL library 126
system-managed buffering
buffer optimization 35
description 35
direct optimized 36
direct optimized buffers 37
direct optimized hiperspace buffers 38
direct weighted 36
___location of buffers 38
LSR resource pools 39
requirements 35
RMODE31 JCL AMP parameter 38
sequential optimized 37
sequential weighted 37

T
tailored compression dictionary token 17
tailored compression, evaluating 14
TAPECOPY processing 53
TGTGDS keyword 60
TTOKEN keyword 31

V
VRSMIN option 92, 93
VSAM
DCOLLECT enhancements 47
DS1REFD 41
enhancements in DFSMS/MVS V1R4 35
fast load implementation 40
last reference date at CLOSE 41
load enhancements 39
RLS KSDS extended addressability 41
system-managed buffering 35
VSAM attributes in data class 44
VSAM attributes in data class 44
migration considerations 47

Index 157
158 DFSMS/MVS V1R4 Technical Guide
ITSO Redbook Evaluation
DFSMS/MVS V1R4 Technical Guide
SG24-4892-00

Your feedback is very important to help us maintain the quality of ITSO redbooks. Please complete this
questionnaire and return it using one of the following methods:
• Use the online evaluation form found at http://www.redbooks.com
• Fax this form to: USA International Access Code + 1 914 432 8264
• Send your comments in an Internet note to [email protected]

Please rate your overall satisfaction with this book using the scale:
(1 = very good, 2 = good, 3 = average, 4 = poor, 5 = very poor)

Overall Satisfaction ____________

Please answer the following questions:

Was this redbook published in time for your needs? Yes____ No____

If no, please explain:

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

What other redbooks would you like to see published?

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

Comments/Suggestions: ( THANK YOU FOR YOUR FEEDBACK! )

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

_____________________________________________________________________________________________________

 Copyright IBM Corp. 1997 159

IBML 

Printed in U.S.A.

SG24-4892-00