?

Log in

No account? Create an account

Previous Entry | Next Entry

Stuff on my other blog

Deadline crunch week rushes on, and in a fit of procrastination I added a couple of pieces to my techblog "A New IT World" (syndicated on LJ as itphasechange):

The Hypervisor Wars and How long before there's an Apple Hypervisor?

Enjoy.

Comments

( 3 comments — Leave a comment )
autopope
Jan. 11th, 2006 07:43 pm (UTC)
Brain dead! Aaagh! IBM had fully self-virtualizing OSs and hardware to support an arbitrary number of VMs back in the 1970s!

This is so obviously a manufactured problem that it's unbelievable. Shoot Intel. Shoot Intel now.
sbisson
Jan. 11th, 2006 07:50 pm (UTC)
Well, IBM did have the advantage of owning the hardware, the OS and the VMM - the VMM being what we now tend to call the hypervisor.

The Intel/AMD virtualisation model is the same but different. They're providing the hardware instructions that allow client OSes to run at Ring 0 where they should (instead of Ring 3, where the current VMM technologies have to run - hence the hideous hacks that VMware and the like have to do when intercepting processor calls). They're just leaving it up to the software industry to deliver the VMMs and hypervisors.

Without a set of open standards for handling images and partitions I can't see any way we could get VMM/hypervisors to cooperate - let alone run at the same time.
ciphergoth
Jan. 11th, 2006 11:37 pm (UTC)
You mean that you can't run a hypervisor on top of a hypervisor? FFS! Then it's not a hypervisor at all, it's just another layer of goop. It's not a hypervisor until a program cannot tell whether it has been given the most privileged level of access on the processor, or whether it has a level of simulation.

Of course, this runs entirely contrary to TPM.
( 3 comments — Leave a comment )