EVA and best practises regarding RAID sets (RSS)

[AWo]

Regarding the EVA best practises a RSS for a disk group starts with the first disk in the first (leftmost) slot of an enclosure which is bay 1 continuing in this column. So the next disk is disk in bay 2 then bay three then bay one of the next enclosure. Regarding this way to create a RSS we were told that it is better to add an enclosure with disks or at least to keep a column complete (a disk in each enclosure).
But when I display the position of our disk groups (regardless of how many disks are member of that disk group) they are spread around all enclosures and it looks like this was done randomly. We're expected to see a clear order of these disks.

Any explanation?

Sincerely
AWo

[Torsten.]

The RSS members are "moving around" when doing group/ungroup followed by merge and split operations.

[AWo]

So what is the explanation if that happens on a clean system where we have created only one disk group? We figured that issue out in a lab session. We created a disk group with 8 and after that with all disks. Same behaviour. After we saw that we had a closer look to our productive installation (3 x 4400, 1 x 6400) and we saw the same behaviour. The 6400 is pretty new, one disk group, created from scratvh, no disks has been changed, yet. Same thing.

AWo

[McCready · Accepted solution]

As stated in another thread, customers are not really supposed to know about RSS, look for it, or care about it. HP even removed an indicator for a good RSS configuration a few versions back from command view - seems they were getting too many worried customer calls.

That said, there is some way (beyond my knowledge) that an EVA will try to lay out disks to form an RSS group, but even if it does it "perfectly" on a brand new EVA with at least 8 shelves, once you add some more disks, shelves or have disks fail and be relaced, all bets are off as to how "nice" the RSS groups will be laid out.

The short story is - why worry? Unless you have a total shelf failure, you are unlikely to lose access to data. And over time, it is almost impossible to expect the EVA, with disk replacements and additions, will keep things laid out nicely preventing at least one or two of your shelves being at risk of failing and causing loss of data access until replacement.

[AWo]

Thanks for that explanation.
I have no worries but I wanted

- just to know it because I'm curious (that's part of my job and my attitude to work)

- to be sure that everything is running correctly as the behaviour we see doesn't fit to what HP wrote in the documentation.

- I want to prevent sub-optimal operation due to a wrong configuration when the box gets expanded as everybody (HP partners, documents, HP teachers emphasized on that when an EVA gets expanded).

But it seems that this won't happen so everything is fine.

Sincerekly & thanks
AWo

[AWo]

Take the "sh" out of IT.