x-point, wavedreamer, thanks very much for responding. I was hoping someone would jump in.
One of the confusing aspects was trying to understand why virtualization would be used if each domain is not secured from the others. I re-listened to the HAP video from September’s NSA conference [link here: http://www.nsa.gov/ia/programs/h_a_p/index.shtml ], and I gather that virtualization was meant to secure the separate domains, but simply doesn’t succeed against a sophisticated attack. .
That video also confirms what x-point said, “What makes HAP special is utilizing special hardware/firmware, as in Intel's VPro the iTPM, TXT and VTd and late launch (DRTM)features, that provide the protections to guarantee the isolation of the environments where these OS's are running.” I also believe they said that not all operating systems can be used if the domains want to be secure.
The video also makes the point that domain isolation cannot be absolutely guaranteed (although I understood them to say they don’t know of a remote attack that could succeed against the HAP set-up), and that is why Trusted Computing's measured boot and host attestation is part of the HAP set-up.
Helpfulbacteria said in the post I was responding to (# 207099), “I do not anticipate Wave making much money off of HAP.” I assume that means that he does not believe HAP will be widely deployed in industry. If that is true, then we are back to the initial difficulty: If the domains are not secure, why deploy virtualization? All that effort just to make it somewhat more difficult to compromise important data?
Thanks again for your help. It is much clearer now.