Hi guys,
We are building an AFF cluster that will provide nfs services only.
The plan is to create a ssd datastore cluter in the primary vc which we will move virtual machines to based on the results of some scripts (IO Density, throughput, latancy and so on).
We are not sure what is the best size for the volumes for this sizes of the aggregates and to build the volumes with qtrees or without?
- Aggr1 - node 1 (8080), size: 125.3TB
- Aggr2 - node 2 (8080), size: 125.3TB
- Aggr3 - node 3 (8060), size: 19.64TB
- Aggr4 - node 3 (8060), size: 13.35TB
- Aggr5 - node 4 (8060), size: 19.64TB
- Aggr6 - node 4 (8060), size: 13.35TB
Disk shelves: 5x800g assigned to node 3
1x15.3t assigned to node 1
Im realy not sure about this architecture, that's why I posted all this information and I realy need some detailed comments.
Another detail, with IOmeter I was able to create 500k Iops on 4k Random workload, pretty low for me.
Thanks in advance!
