That exact problem was why security kernels and later separation kernels were invented. The theory was you could bulletproof 4-12Kloc of mostly static code more easily than monolithic, dynamic hypervisors. They have less nifty features, though. Really bare bones.
Could someone more knowledgeable here explain this to me? How do you "crack the password using ssh" from a guest VM to the host? That seems to imply that the guest VM and the host were somehow on the same network and the host had a SSH service. Is this the case? And if so is that normal (or the default?) config, that you can network from the guest VM to the host?
> Ran mysqldump against the production database, creating 1.txt
Hack 2:
> Ran mysqldump against the production database, creating 1.txt
Anybody else sees a pattern here? We need to forbid creating files named "1.txt"!