Aggregates Capacity Used Above 96% & 96% of IOPS on our arrays falls on 20% of our capacity

[Bhuppi]

Hello. We've four out of 8 Aggregates at 96%.  I believe volumes inside them are thick provisioned as well. Not sure how to check and confirm that thick/thin part. But I know this is dangerous and performance issues may occur. How to take care of this. Can we move around between aggregates or add to an aggregate from another one directly to increase space.  Any capacity planning doc on Aggregates over 80% utilization. Also, want to know how aggregates are formed and why they are formed the way they are, by folks who're managing it earlier. 

2nd Part of question: How to check and make sure flash cache is turned on and configured correctly on applicable nodes, and any straightforward command to identify unwanted snapshots and volumes which are not mapped to any hosts.

Finally, I've been told that 96% of IOPS on our arrays falls on 20% of our capacity (54 TiB). What does this entail and can this workload concentration can be properly distributed. Any technical doc on this as well.

Thanks in advance.

[dbenadib · Accepted solution]

Hi Bhuppi,

How to check thin / thick volumes:

      vol show -fields space-guarantee

Volume = Thick Volumes

None = Thin Volumes

Can we move around between aggregates or add to an aggregate from another one directly to increase space?

Yes you can move non disruptively volumes across aggregates. If you want to spread a volume across aggregates then you need to create them as a FlexGroup (https://www.netapp.com/pdf.html?item=/media/7337-tr4557.pdf)

command to identify unwanted snapshots and volumes which are not mapped to any hosts.

you can list snapshots and their created dates using the command:

• set d -c off; snapshot show -fields create-time,size -sort-by create-time

you can check volume that are not mounted (not able to be accessible by any host) using the commmand

• vol show -junction-path -

Another idea would have be to look into the export-policy to check if hosts are still running 

• vol show -field policy
• export-policy rule show -policy name_of_the_policy

HTH

David B