2016-12-05 07:22 AM
I run a FAS3220 cluster, running the 8.2.1 version, 7 mode.
I have one aggregate aggr0. This aggregate contains around 10 volumes.
On the other side, this aggregate is composed of 2 raid groups, of each 23 disks, in raid DP.
here is the issue I have...
some users encounter lag & performance issues.
When I checked disk activity, I saw that the disks of the 1st raid group are almost allways over busy, with IOPS almost at the limit for that kind of disk,
and on the other side, the disks of the 2nd raid group, are almost unused! (around 10 - 15IOPS average...)
So I guess that there is an issue with the volumes distribution along the aggregate and the Raid groups?
I identified the 3 volumes the more "IOPS-greedy"
Is there a way to better distribute these 3 volumes along the 2 Raid groups? Would a volume reallocate make that? Or is there maybe an other way to do that?
Thanks a lot for your help!
2016-12-05 07:28 AM
It seems that when the second rg was added, there was no reallocate done so the data can be spread across the new spindles.
For 7-mode, look into the Re-allocation commands. Google for this TR - reallocate best practices tr-3929