New C80 Storage Aggregates

[marshit]

Hello All,

Quick question. We all know with the newer storage arrays they have ADP already enabled so you don't need to have as many spare drives as traditionally you might have before ADP.

If you had a new C80 delivered to your door step, all SSD, 24 drives, RAID-DP which would be a better scenario or more in line with NetApp recommendations.

I know much of the decision depends on workload and other factors but It seem like option 2 is the better choice.

Create 1 big aggregate using all usable partitions

• ~66 data partitions (from ~22 drives)

• RAID-DP (12–14 drives per RAID group, automatic grouping by ONTAP)

• 1 spare full drive (3 partitions)

Split into 2 aggregates, each made of two RAID groups.

• Each aggregate = 33 partitions (11 data + 2 parity) × 2 groups

• Leaves 1 full spare drive

Split into 2 aggregates but with smaller RAID groups (10 data + 2 parity).

• RAID groups of 12 partitions (smaller size)

• Each aggregate ~2 RAID groups

• Leaves 1 full spare disk

[TMACMD · Accepted solution]

Please keep this config as two aggregates. Each aggregate can have up to 23 partitions (1 spare partition) depending on risk factors. This is typically the default. Many customers prefer to leave two spare partitions, so a raid-group of 22 partitions with 20 data and 2 parity per node.

This allows both nodes to be active. If you assign all data partitions to one node, you essentially have an active/passive config. One node does all the work (at least data). The other node can handle some network traffic, but will ultimately pass the traffic to the node with the storage.

Keep you performance. Keep it active/active.