When you took the VMware-based snapshot of the VM on your 'Echo Mary Charlie' array, was it the *exact* same VM that's running on the Nimble? If you have a lot more memory access going on with the Nimble-stored VM (say, it's a file server servicing many inbound clients, reading and writing) and, say, a different VM (that's idle) on your 'Charlie array, I'd expect that the VMware-based snapshot process would run *way* slowerm, as it has to quiesce the VM and write out the memory contents. If you didn't storage vMotion the same VM between the two arrays, I'm afraid it's not quite a fair comparison! Note: I'm basing some of my commentary from a similar thread found on VMware's community site: Why are snapshots with memory so slow? | VMware Communities
So....apples to apples, or apples to oranges?
Thanks for your patience, Ron!
I think it was a fair comparison that I tried on vm's of various sizes none of them loaded with user sessions. Note quiesing is an optional component of a snapshot and one that I did not include. The VNX is at capacity and running most of our environment while the Nimble has next to nothing on it. So just to be sure I just did a snap of a small 2GRAM/40Gig of storage vm on the VNX. 45 seconds with memory. delete snap and svmo to the Nimble... the time was quicker.. 25 seconds. Hmmm. So try a larger 8GRAM/600 gig of storage. 6 minutes to snap with mem on the VNx. Move it to the Nimble... and 6 minutes to snap.
The difference between two days is I've move little more load onto the Nimble. Must have been cache warm up
Note to moderator, you can delete this thread if the title appears disparaging.