Subscribe

Reallocate on Snaplock aggregate after disk addition

We are expanding our Snaplocked aggregates by a couple of disks, and are trying to avoid creating "hot disks".

Can you run reallocate on Snaplock aggregates? Would there be any benefits?

I think it would be similar to running reallocate on volumes with snapshots, which means very little benefits. Has anyone

done this and seen results?

Thanks.

Re: Reallocate on Snaplock aggregate after disk addition

Hey Fabian,

(I cannot comment on the Snaplock part)

I have worked on filers which were under tremendous load, and every once of performance we could squeeze out of them was important.

Yes I have done WAFL reallocate on Trad Vols in the old days, which are essentially what AGGR's are today.

Yes I have seen performance improvement and the elimination of "hot disks", I would recommend it based on the following constraints.

1.     You always will have a chance of data loss running this command, so make sure your management is aware of this factor. (backups, risk mitigation, etc)

2.     You really need the faster performance for the environment, vs a like too have.

Hope this helps.

Anthony Feigl

Re: Reallocate on Snaplock aggregate after disk addition

Hey Anthony,

We currently have reallocate scheduled and running on all our volumes that don't contain Snapshots.

I see the improvements on these specific volumes, but with Snaplock, we are unsure if we can

eliminate these hot disks. I've searched the NOW site and have found nothing. Are you aware of any

documentation detailing reallocate on snaplock volumes/aggregates?

Thanks for the help

Re: Reallocate on Snaplock aggregate after disk addition

Hi Fabian,

Currently the reallocate command isn't supported on SnapLock aggregates.

Thanks,

S.Naresh

Re: Reallocate on Snaplock aggregate after disk addition

To my knowledge its never been best practice to add a couple of disks at the time. for the exact reason you mention; avoid hot spots.

Add many disks and you ll be fine. i know this answer does not sit well with those that pay the bill for the disks but hey... thats the answer

anyways to avoid degraded performance.