Thanks, snapshots -> I/O overhead is well understood. I was rather wondering whether ESX can put a measurable figure on the very specific I/O overhead that is occuring right now due to snapshots. Or at least some indication that one is bordering on dangerous dimensions.
In our case we consolidated snapshots in order to move VMs to another datastore, not because we suspected them to be the primary reason for our far-below-average I/O performance. If there was an indicator that said "watch out, too many snapshots, too much I/O overhead" this could speed up the resolution of snapshot-induced I/O overhead issues. Or at least bring better attention to such kind of issues for everyone.
Sebastian
