I do/have all dc’s vm’ed in vm environments. I do understand the ‘feel good’ scenario. However, I place as much trust in the esx delivery as I do the physical box, if not more so since the vm is portable. That’s my .02. This structure is based around a san with or w/o vmotion (depending on budget). You can put a san together for next to nothing, that is fairly robust (drbd/openfiler/iet). With Esxi, and no vmotion. If a physical esx goes down you simply start them on the other machine. Its not pretty but it works, and a small/med shop that can be down for 15 minutes while someone does that can be worth saving the 5k. If you are talking about putting all your eggs in one basket that’s acceptable in small/med environments. I know you mention this is a large environment so then I wonder how big could it be with just 1 exchange and 2 dcs ? Everything regarding restoring vm’s with DC data is well documented and there should be no surprises for those who need to do that. On the physical box, if the server crashes you have it even worse, because you cant just ‘restore’ the image. You have to go rebuild the server, join the domain and then promote it again, or restore from backup. That actually sounds like more work, then restoring the vm, rolling back the ticket. In your particular case, it sounds like the advantages vmware gives you , space, consumption, failover have all been swept aside to save dollars. Therefore, I would say you should rethink the current solution.
~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/> ~
