Best Practices for HP EVA disk groups

[Kites BM]

Hi,

We are working on HP EVA 4400 array.

May I know the best practices for disk groups creation including the vdisks,parity and hot spares for future expansions.

Presently we have 2 disk groups with 23 vdisks each,

please suggest the best practices for future expansions.

[IBaltay]

Hi,
this best practices are for other EVAs, but most of this is valid for all EVAs:
http://h20195.www2.hp.com/V2/GetPDF.aspx/4AA0-2787ENW.pdf

[Kites BM]

Presently we have 2 disk groups with 23 vdisks each,

i referred the white paper on http://h20195.www2.hp.com/V2/GetPDF.aspx/4AA2-0914ENW.pdf ,

please let me know

1)if we have to add a disk each in one group
or
2)how do we fit the 23 disks in each group to match "disks in multiples of 8" - (this includes parity and hot spare as well) ( do we add a disk each in each group to match the count to 24 disks(8 * 3)? or resize the available disks to reduce them to 16 in each group?)

[IBaltay]

Hi,
internaly the disk group is subdivided in the so call redundant storage sets (RSS), which default is 8 physical disks. Good idea is to check the current RSS state (numbers of disks in each existing RSS) and if you have e.g. RSS1=8disks, RSS2=8disks, RSS3=7disks, then add/group one additional disk to the RSS3 to have 8 too...

[Uwe Zessin]

I agree to add one disk to each group to upgrade to 24 disks each.

You've been asking about 'parity' + 'hot spares'...
Well, the EVA's virtualization automatically distributes the capacity across the disk group. And despite the first "R" in RSS, an RSS does not provide any redundacy. (Don't blame IBaltay for it - I don't think he was responsible for the misleading name ;-)

The recommendation is to set the so-called "protection level" to single. In reality it does not 'protect' your data - it is the virtualized equivalent of 0,2 or 4 spare disks.

The real protection is provided by a virtual disk's VRAID-level (1,5,6). VRAID-0, like a traditional RAID-0 DOES NOT provide any redundancy.

There must be at least 5 GigaBytes of free space in the disk group if you don't use other features like snapshots or replication, but the recommendation is to keep free the capacity of another two disks.

The documentation does some techno-babble about a "PDM" (Proactive disk management) request. The idea is that there is still capacity left if the EVA or the end-used removes a disk drive from the group for maintenance reasons.