I have been trying to ‘fix’ the issues I had discussed in the original posting with our servers which had undergone a Physical to Virtual migration.
So far I’ve tried two things:
- Disable Symantec Antivirus
- Switch the HAL of the server from ACPI Multiprocessor to ACPI Uniprocessor
I happened across the second option through some google searches. Have a look at the following URLs:
http://kb.vmware.com/KanisaPlatform/Publishing/647/1077_f.SAL_Public.html
http://www.vmware.com/community/thread.jspa?messageID=415307
http://www.vmware.com/community/thread.jspa?threadID=29202
It’s only been 1 day so far, but our wwoods-dc2 server I had talked about before is now humming in the 0-25% CPU utilization range. This seems to be a huge difference from what it was running at previously. Disabling Symantec Antivirus wasn’t noticeable.
Update (10/30): I’m still running in the range of about 15% CPU utilization average on that server. I think I should be clear that it wasn’t the utilization of the VM that was the issue, it was that CPU utilization was erratic instead of determined by the true load of the server. I think this is another case of newbies to Virtualization not being privy to information that is obvious to the old hands. This is similar to the ESX & IDE harddrive issue I posted about earlier.
I’ve received a comment from David at vmblog.com suggesting removing vendor drivers or utilities. When we first migrated these servers, I did my best removing any HP utilities and services that I could. I’m sure there is more work to be done.
I found a CD called p2v-scripts.iso on the VMTN forums. Link below. I’m going to give it a shot when I get a chance to shutdown the server. Essentially, it’s an automatic way to remove unnecessary drivers from a P2V’d VM. There are other goodies as well. Thanks to Mike for putting it together.