Redundant Set of Storage question

[Timo Steylen]

On an EVA5000 from one of our customers i noticed the following RSS sets (RSS-index / amount of disks):

1/10, 2/8, 3/8, 4/8, 5/8, 6/6, 7/8, 8/8, 9/7, 10/8, 1/8, 12/8, 13/8, 14/8, 15/7, 16/8, 17/6, 18/10

Does it matter that some RSS-sets have an odd amount of disks (what about mirroring/widow disks)? And some RSS-sets has 10 and some has 6 disks?
Is it possible to start the releveling of disks to get as much as possible RSS-sets of 8 disks?

Are RSS-sets created within each diskgroup or is the data from different diskgroups spread over all (and sometimes the same) RSS sets?

many thanks!

[Patrick Terlisten · Accepted solution]

Hello,

AFAIK is in RSS with an odd number of disks, one disk unused. Every disk in an EVA has a partner disk in the same RSS. If the number of disks is odd, one disks doesn't has a partner and it's ununsed.

Hope this helps,

Best regards,
Patrick

[Uwe Zessin]

> it's ununsed.

It is unused for VRAID-1 data, but VRAID-0 and VRAID-5 data can be stored on such a disk drive.

[Víctor Cespón]

This must be a transient state after a disk group member change. The rules are that there can be only one RSS with an odd number of disks. After a disk is added or removed to a disk group, you can end up in an unusual configuration, but soon controllers will start merge/split operations.
A RSS is a subgroup of disks inside a disk group. Information is divided in parts and each part stored, with the RAID overhead, on a RSS.
For the data in vRAID 1, an even number of disks are used, if the RSS has an odd number of disks or disks of different sizes, not all available space will be used.

[Patrick Terlisten]

Ah, okay. I thought that in RSS, with an odd amount of disks one disk, one disk always is unused, even if VRaid 5 oder VRaid 0 is used. THX for clearness.

Best regards,
Patrick

[Timo Steylen]

ok thanks all for the response!

i was assuming that you can (somehow)manually regroup the RRS sets...

but how is it possible that 2 groups has had a odd amount of disks and some groups has 6 or 10 disks? does it have influence on the performance?

maby a new firmware (from 3.028 to 3.110) will resolve this..