Examples of changing the default startup priority to high would be;
Infrastructure servers - AD, DNS, VC, WINS, DHCP
Application wise - SQL backends being brought up prior to web front ends etc.
Changing the default response to leave powered on, there was some talk on the boards around this being the standard approch to take. The primary reason for powering down VMs in the event of HA is to release reservations and locks on the VM, allowing it to be registered on an alternate host. If a host has gone down hard, then if you think about it the locks are going to be released anyways. During a HA operation, my understanding is that the HA cluster will (via primary nodes I assume) will check for these locks if they don't exist the servers will be powered on alternate hosts, if they do exist then they will not. It is not as cut and dried as reading the deafult response file and following orders as it were.
So by having an isolation response of stay powered on you are not neccissarly losing the ability to power them onto alternate hosts in the event of a machine down faliure. There may be a issue if there was a network outage on one host only affecting both COS and vmdata networks.
Simon
https://virtual-simon.co.uk/
