ONTAP Hardware
ONTAP Hardware
We have a FAS8040 (Clustered DataOntap 9.1) that currently has the following configuration:
aggr1_01 - 20 disks (18+2 raid dp) - node1
aggr1_02 - 44 disks (two 22 disk raid groups (20+2 raid dp) - node2
I am adding a new shelf of 24 disks with the intention of adding all 24 disks to aggr1_01 to address performance and capacity issues.
Can you please look over my steps below and let me know if I am missing anything or simply confirm what I am doing is correct?
At this point I should have successfully added 24 new disks to aggr1_01 and it should consist of two 22 disk raid groups, just like the other aggregate aggr1_02.
In order to take advantage of the additional spindle count, do I need to reallocate all the volumes on aggr1_01??
What am I missing or not considering?
Thanks for your help!
Looks good.
Are you good? I saw the same post on reddit.
The only remaining question is regarding my raid groups.
I was initially planning on adding 2 disks to the existing 18+2 raid group to make it a 22 disk raid group and then add the additional 22 disks to the aggregate as a second 20+2 raid group. This would leave me with a 44 disk aggregate consisting of two 22+2 raid groups.
My concern is by adding those 2 disks to the existing 18+2 rg, will that create an imbalance/hot disks? Would it be just as good to leave the 18+2 rg alone and simply add a new 22+2 rg to the aggregate?
I will be running reallocation jobs on all the volumes after.
Match your RG as best as possible. a difference of 1 or 2 disks rarely makes a notable difference, but anything more than that I would adjust to they are close to the same, if not the same.
If you have enough spares, I would create 2x 22 disk RGs. Any "hotspots" should be taken care of after the reallocate is done. It will move the blocks over any new disks added. Even if you don't run the reallocate, WAFL will eventually use the new disks for writing date. the reallocate just speeds up the re-distribution of blocks to new disks in the aggr.