AO, growth warnings and Snapshots

[P1ct0]

Hi,

So I was wondering what the best practice is for setting growth warnings on NL R6 CPG's. My understanding is that if you have thin provisioned VV's within a non AO CPG on NL disk, you must ensure that your growth warnings are set to allow for the space that you could use up in your Non AO CPG.

To avoid this problem, the non AO VV's, I converted them to thick provisioned.

However, I'm left with the snapshots which vary in size, and have their own CPG which is not part of AO. How do I know what to set the NL AO CPG's growth warnings at if im not sure how large the snpashots might be?

Hope this makes sense,

Cheers,
Shaun

[hdtvguy]

Win no AO th warning is simply a warning that will send you an alert to let you know when you hit that number. In AO it is used to steer AO from placing any new blocks in that CPG once the limit has been reached.

[P1ct0]

Win no AO th warning is simply a warning that will send you an alert to let you know when you hit that number. In AO it is used to steer AO from placing any new blocks in that CPG once the limit has been reached.

Ah, I may have not been very clear sorry. for instance, I have 4 NL CPGs, one is Non AO 6x 2TB of space allocated in thin provisioned luns on NL, another is Non AO where my snapshots are, and the other 2 are in AO. I have 57TB of total NL disk.

The question is, am I correct in having my Snapshots in a non AO CPG, and if so, what would I set my growth warnings at?

Thanks

[hdtvguy]

I put all our snap shots in a non-AO tier. As for all the CPGs in NL you have I would caution that if you run your array lean you may run into issue as one tier may run out of space and if the array does not have truly free space you can have issues. Just because another CPG has free space if the system NL does not have enough free space you can get into trouble. We used to run 3 CPGs in each disk tier for various apps and realized as the system got tight on space this was creating issues. Since AO is not all it is cracked up to be and slow at reacting we are moving to a single AO policy per array and just 5 CPGs, 3 for AO, 1 for snaps and then one NL that we pin all our low IO files systems to.