Need some advice on aggregate creation

HUX20002000

I'm configuring a NetApp FAS8300 for use specifically as a backup target. It will likely host only one VM (Commvault) that will connect to some large volumes for data storage. I have the following shelves/disks/ideas for configuration:

• Shelf 1: 6 x 900 GB SSD, create 2 x storage pools (3 disks each = allocation unit size 671 GB; apply half to the shelf 2 aggr and the other half to the shelf 3 aggr)
• Shelf 2: 12 x 4 TB SAS, create 1 x aggregate (1 x RAID-DP group of 11 disks, 1 spare =  30 TB; host the Commvault VM here)
• Shelf 3: 60 x 4 TB NL-SAS, create 1 x aggregate (2 x RAID-TEC groups of 29 disks, 2 spares = 179 TB)

This would result in an active-passive system, using only one node's CPU capacity. 

But then I was wondering: does backup storage even benefit from being backed by an SSD storage pool? Probably not much if at all, right? If that's case, maybe this would be better:

• Shelf 1: 6 x 900 GB SSD, create 1 x storage pools (6 disks = allocation unit size 1.3 TB; apply all to the shelf 2 aggr)
• Shelf 2: [same as above]
• Shelf 3: 60 x 4 TB NL-SAS, create 2 x aggregate (1 x RAID-TEC groups of 29 disks each, 2 spares = 90 TB per aggr)

This would result in an active-active system, using both nodes' CPU capacity.

Or perhaps a third option:

• Shelf 1: 6 x 900 GB SSD, create 1 x aggregate (1 x RAID-DP group of 5 disks, 1 spare = 2.4 TB; host the Commvault VM here)
• Shelf 2: [don't use it]
• Shelf 3: [either 1 or 2 aggrs, as above]

Any opinions on this?

TMACMD

Note, commands are close. All by memory. May need to tab-out and correct as needed. the gist is there.

Connect all drives.

Stack 1: DS460C, then DS212C

Stack 2: DS224C

Re-init, 9a and 9b

(this should maximize the ADP (i.e. minimize root partitions to 12 each) and split across both controllers.

Create a Storage Pool with 5 of the 6 SSDs

Assign all SSD to one node

storage pool create -pool sp01 -diskcount 5 (I think thats correct)

Create two aggregates, one per controller. You may need to do this manually to maximize capacity

You should be able to do something like

aggr create -aggregate sata_01 -node node-01 -maxraidsize 14 -diskcount 28 -is-hybrid-enabled true

aggr create -aggregate sata_02 -node node-02 -maxraidsize 14 -diskcount 28 -is-hybrid-enabled true

If that fails, make the aggr:

aggr create -aggregate sata_01 -node node-01 -maxraidsize 14 -disklist 1.0.1,1.0.2,... -> specify all 12 partitioned drives for this node -is-hybrid-enabled true

aggr create -aggregate sata_01 -node node-02 -maxraidsize 14 -disklist 1.0.24,1.0.25,... -> specify all 12 partitioned drives for this node -is-hybrid-enabled true

Then expand:

aggr add-disks -aggregate  sata_01 -diskcount 16

aggr add-disks -aggregate  sata_02 -diskcount 16

You should end up with two aggrs of 82.53TiB each

aggr add-disks -aggregate sata_01 -pool sp01 -units 2 -raid-type raid4

aggr add-disks -aggregate sata_02 -pool sp01 -units 2 -raid-type raid4

Create a flexgroup. Maybe in this case something like this:

vol create -volume fg01 -junction-path /fg01 -aggr-list sata_01,sata_02 -aggr-list-multiplier 1 -size 160T

That should make a flexgroup with two members at 80T each.

Alternatively:

vol create -volume fg01 -junction-path /fg01 -aggr-list sata_01,sata_02 -aggr-list-multiplier 2 -size 160T

Which makes 4 members at 40T each