ADP configuration on AFF C250

[Zulhadi]

Hi,

Just for asking for any expert here. For the ADP configuration that configured from factory is root-data-data-partitioning on the AFF C250 or any AFF system, is it possible to change it to root-data partitioning like the FAS entry level such as FAS 8200, etc.? Thanks

[TMACMD]

You really do not want to do that. WIth ADPv2 on SSD/NVMe you essentially can evenly distribute storage across your controllers and minimize the parity tax.

Example with ADP

Each node must have a ROOT aggr and when possibile suggested to have a data aggr.

With ADPv1, disks are partitioned. half the disks are owned by node 1 and the other by node 2

When I go to create a data aggregate, I use HALF the disks, minus 1 for spare, and typically need to reserver 2 data partitions for parity...on EACH node. I now have 4 parity drives

With ADPv1, disks are partitioned. half the disks are owned by node 1 and the other by node 2

But now, I have two equal sized data partitions. One nodes owns all P1 partitions and the other node owns all P2 partitions. When the data aggregate is created, I can span over the drives, usually use raid-dp and only need two parity partitions.

Effectively, I would be using two drives for parity (two partitions on each node)

Trust me, trust everyone else that comments, stick with ADPv2 on any SSD/NVMe platform

[Zulhadi]

Hi @TMACMD ,

Very appreciate with your response. 

but with the ADPv2, is it possible if we create the only 1 aggregate? or it must have 2 aggregate since it all disk equally to all node (for data partition)?

[SpindleNinja]

You can create a single aggr,  you just end up with a single aggr with 2 different raid groups.  

Why do you need a single aggr ? 

[Zulhadi]

hi @SpindleNinja ,

The idea is to populated all the capacity available from the disk to single aggregate. 

Actually I don't noticed about the AFF Series have an ADPv2 configuration.