2010-02-07 11:02 AM
I am working on an environment where the volumes from Netapp storage are exported to ESX Clusters and these clusters are housing the VDI farms and often I come across some weird scenarios where there's a sudden rise in the number of IOPS and increase in the Latency. Surprisingly this again comes back to normal after a while.
I found that the network utilization is fine during this time, upon collecting nfs per client statistics came across few NFS heavy hitters but those systems are not consitently showing up on executing the "nfsstat -l"
And during the high rise in latency and number of nfs iops the vdi users are barely able to access or login to the Virtual desktops.
I've also noticed that during this time the dedupe is active and the clients which gets listed by executing "nfsstat -l", the dedupe is also active in these volumes exported to the clients getting listed as the nfs heavy hitters
Did anyone come across scenarios as such, if so please share your experiences I am not that good at VMware so if someone can throw some light into this it'll be great.
2010-02-07 10:41 PM
Please have a look at this link http://media.netapp.com/documents/tr-3428.pdf ,this might be of interest to you ,this might help you in identifying or checking environment is set in a optimised way.
2010-02-08 06:21 AM
Hi Sam & welcome to the forums!
ONTAP de-duplication is post-process & does require some additional resources whilst doing scan - so high-level recommendation: try to schedule scan outside of peak hours.
Have a look at this TR-3505 (a.k.a. De-Dupe Bible , just refreshed):