The question I was asked: Can I have two pools in the SSP for AIX rootvg disks and data volume group disks?
Answer: no

Explanation: Only one pool per Shared Storage Pool.

• Alternative is you that could use two SSP Tiers in a single SSP!  Tiers provides more than multiple pool in functions and usability.

Question: Is two tiers a good idea?

Answer: no

Explanation:

• Tiers have separate LUNs for each tier.  Having two tiers to split these types of LU (SSP virtual disks) use is rather pointless and will reduce performance.
• You might end up having to move LUNs between tiers (doable but simply extra rather system admin pointless work).
• One of the strengths of SSP is the humble one disk virtual machine has many LUNs at the back end to provide high speed concurrent disk I/O as the single LU's  blocks are spread across all the back end LUNs.

Below at the top we have vSCSI or the NPIV model - every LU is mapped to a single LUN but at the bottom the SSP is mapped to ALL LUNs:

Splitting the SSP LUNs in to two tiers for rootvg disks and other disks would reduce the natural load balancing to  subset of disks and probably increase system admin workload.  For single disk VMs, I think SSP could out performance NPIV considerably.

Question: So what should we do to identify rootvg LU virtual disks?

Answer: KISS - identify rootvg and boot LU virtual disks by adding "rootvg" & “boot" to the end of the name.

Question: What are the right reasons for multiple tiers?

Answer:  I think there are six? - Please let me know if you think of more reasons.

By example:

1. Fast, medium, slow - to highlight the LUNs have different performance
2. IBM, HDS, EMC - to remind yourself of the underlying disk unit vendor and implications of that
3. Prod, in-house, test - to make sure you know different policies are in place for the important data
4. V7000a, V7000b - so you know explicitly which disk unit the data is in to reduce risks
5. Datacentre1, datecentre2 - to clarify where (geographically) the LUNs are placed
6. Mirroring - to separate disk space that is remotely mirrored (failgrp in SSP terms), locally mirrored or not mirrored at all.