I remember reading somewhere on a way or script which you can run which helps prioritize and retry the boot of critical VMs in a large VMWare farm. It has been a number of months sence I saw the article / post and can't find it.
Infrastructure:
28server VMWare HA farm all working with single large SAN system
HA / VMotion / DRS working and tested on all systems
1 vCenter server running as a VM
Issue:
SAN crashed and all the servers had to reboot but none of the VMs started because the vCenter server was not online. I had to spend about 30min playing "where is my vCenter server"so it could be up and running so it could start telling the 28 servers which VMs to start.
Goal:
To set the host running vCenter such that it will boot upon completion of the ESX server boot cycle and if the host who last owned the VM is NOT online that some host in the farm will grab and boot it.
With customer's having larger and larger VMware infrastructures, I have to believe that this is taken care of in an automatic way.