2011-08-19 03:14 AM
We think we are seeing disk throughput bottlenecks on one of our aggregates, via the statit command, we have a second aggregate that is doing next to nothing.
The bust aggregate is spilt into 3 raid groups one is slightly smaller than the others so we are seeing slightly higer utilistation on those disks, though in general it is pretty high.
I understand that flexvols are spread across the aggregate but is there a way of seeing which volumes are spread across particular phsical disks in the raid groups?
This is just so that we can try and move those first.
Any tips anyone has would be much apreciated.
2011-08-20 06:46 AM
aborzenkov wrote:is there a way of seeing which volumes are spread across particular phsical disks in the raid groups?
No. You have no control over physical data placement.
Well, that's entirely true. You could use the reallocate command to see if data form all volumes (and the aggregate itself) is spread evenly across all disks in the aggregate, though you can't see the layout. Reallocate allows you then to spread the data across the disks in the aggregate. This could significantly increase performance, but it doesn't work in all situations.
Hope it helps,