IBM I and Z (each with a dizzying array of branding zigs and zags going back decades) are both pretty interesting technologically.
I has a fancy memory architecture, very smart disk controllers (essentially distributed intelligence, like an octopus), a virtual instruction set (that has been used multiple times to almost seamlessly jump huge under-the-hood processor changes), and historically a reliability record second to none (the old box in the wiring closet running for years upon years, completely untended). Z has even more toys, including some of the strongest clustering, partitioning, security found anywhere. Sysplex, LPARs, and RACF are all impressive, especially given how many decades ago they started. We won't even talk about the DBMS and transaction monitors, which are their own brand of crazy strong.
Those immersed in the higher-volume, standard microprocessor, Unix/Linux or Windows, cloud mainstream don't give "proprietary systems" much thought or respect. But we probably should. Those who knew the IBM I or Z, or the DEC VAX/VMS, HP MPE, Tandem NonStop, etc.—they were too expensive, too few in number, too quirky—but what they did well, they did outstandingly well in their purpose-focused, allopatrically speciated ways. Better in many cases that we can do today with the latest 2024 gear.
I think this is the biggest problem, plus the fact these systems tend to be tied to proprietary - and also very expensive - hardware platforms. If I want to learn about GNU/Linux or BSD, all I need is a computer (PC in most cases, but other options exist) and an Internet connection. These days, most people (at least in Europe and North America) have these anyway, so it's really easy to get started in the comfort of one's own home.
Having a free account on a public machine is cool, but it's not the same as having your own system, especially if you want to learn about system administration.
The killer, of course. As they say: anyone can build a bridge that stands, but it takes an engineer to build a bridge that barely stands. In this game, a solution that's too expensive is often not a solution at all.
I has a fancy memory architecture, very smart disk controllers (essentially distributed intelligence, like an octopus), a virtual instruction set (that has been used multiple times to almost seamlessly jump huge under-the-hood processor changes), and historically a reliability record second to none (the old box in the wiring closet running for years upon years, completely untended). Z has even more toys, including some of the strongest clustering, partitioning, security found anywhere. Sysplex, LPARs, and RACF are all impressive, especially given how many decades ago they started. We won't even talk about the DBMS and transaction monitors, which are their own brand of crazy strong.
Those immersed in the higher-volume, standard microprocessor, Unix/Linux or Windows, cloud mainstream don't give "proprietary systems" much thought or respect. But we probably should. Those who knew the IBM I or Z, or the DEC VAX/VMS, HP MPE, Tandem NonStop, etc.—they were too expensive, too few in number, too quirky—but what they did well, they did outstandingly well in their purpose-focused, allopatrically speciated ways. Better in many cases that we can do today with the latest 2024 gear.