Impressions about snapshot removal in vSphere 6

[tinto1970]

Hi folks,

as discussed in Gostev's weekly email, the new snapshot removal process introduces in vSphere 6 is a great change. And it so good in reducing stuns on VMs during the process.

But, it's just an impression... (if i knew about the change i would have tried to pick up some numeric data "before and after" upgrade) i've noticed that the process has greater impact on storage systems holding VMs data.

In particular case, i have some NFS datastores on NetApp filers which are heavily loaded: when removing snapshots, the IOPS increase a lot and the CPU usage grows up quite high, much more than before.
I.E.: a VM which is around 600 GB, approx 8 hours snapshot had this activity during removal: cpu% on storage, iops and latencies on datastore

[dellock6]

Actually, we measured the time needed to be completed, which is the main cause of the stun as the vm is not responding for a longer time.
Just my guess, maybe the process now is more efficient so it's able to push the storage more, thus the values you are seeing.

Luca

[tinto1970]

yeah, better to consider this when buying new hardware