HPE Storage Users Group

Hello

We have a 2 node v400 with 4 cages. CPG's are setup raid5 (3+1) with the 'cage' availability. When a tpvv gets created from the cpg the status shows availability as being 'port'.

I understand and have tested the fact the 3par provides the best possible spread of chunklets and tries to get to the most resilient layout.

My question is how is port more resilient than cage? In what circumstances can port availability help me out in the event of a hardware failure that cage does not? I assume by port pair they mean the two ports that my cage connects to?

The man pages say the following:

avail. Indicates availability characteristics associated with a created LD. Availability determines from where space chunklets can be allocated when one of the LD's chunklets fails. Availability characteristics are as follows:
- disk. Chunklets in the same RAID set may reside on the same disk.
- mag. Chunklets from another disk within the same drive magazine can be used as a replacement.
- cage. Chunklets in the same RAID set belong to disks on different cages.
- port. Chunklets in the same RAID set belong to disks on different port pairs.
- ch. No redundancy is provided for the logical disk when a chunklet fails. This is only valid for RAID-0 LDs.

This seemed to be something which was introduced in InformOS 3.1.1, we've logged a call with HP about this.

Their explanation was not to our satisfaction. It said that the availability will never go below what is configured, but in this case the GUI shows port availability because that implied cage availability according to our contact.

In our opinion it's the other way around and we're still discussing this with HP.

See this post:

viewtopic.php?f=18&t=107

I had the same issue and was not satisfied with the answer.

I liked the answer they provided. They said PORT availability vs CAGE availability applies to older systems that daisy chained multiple cages off a single port. T400/800 do not daisy chain, and each cage direct connects to a port, thus the net result in a T400/800 configuration is that CAGE avail also meets the requirements for the old school PORT avail, due to the I cage per port.

Setting a CPG to cage availability follows the rules defined by cage, but since each cage has its own dedicated ports that are not shared with any other cages, you're also simultaneously meeting the requirements for the higher grade PORT AVAILABILITY definition.

Perhaps the term "LOOP Availability" would have been a better name for port availability.

Yes. I agree. Would just have liked someone to tell us that. For a V class port == cage.
Thanks for the help.