So recently we came across a performance issue with a bunch of VM’s on one of our clusters. Turns out from doing so much digging that they all shared a common thread, they were part of the same resource pool. The net effect was the VM’s contained were being constrained for CPU over their non-resource pool counterparts(the reason for the sudden CPU spike was as a result of a dodgy AV update being rolled out across the VM’s).
Removal of the resource pool helped settle things down somewhat (+ removal of said DAT :)). Reminded me of a particular slide from the excellent VMworld session which discussed the vey subject (http://www.youtube.com/watch?v=6qg7pVNuAnU).
As noted above in our environment we had 12 VM’s essentially competing for the same level of resources as a single VM in the root resource pool. This would only because an issue when resources started becoming scarce and the resource pool share values would kick in