Vraid 0 Redundancy

[paternostertje]

Does the EVA 3000 provide any redundancy on a Vraid0 Vdisk? Does the level of disk failure (set on the diskgroup) protection provide any redundancy on a Vraid0 Vdisk?

[Andrzej Kowalik · Accepted solution]

No, see:
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=837300

[paternostertje]

tnx. I wasn't sure, now I am. Points are assigned.

[paternostertje]

just to ba sure ... The level of disk failure protection applies to Vraid1 and Vraid5?

[Andrzej Kowalik]

Yes

[Andrzej Kowalik]

its works like "spare space" and consume 0, 2 or 4 x space of biggest hdd in disk group; but Vraid have to have possibility to "survive" disk failure -> only vraid1 and vraid5

[Madis Tuulparg]

Hi,

I did understand, that even v0 can survive EVA disk failure. But only then it has no striped data on that disk. Is it right?

[SAKET_5]

Madis & Doom,

No - Any Vraid 0 LUNs are not protected if any member disk fails.

Disk Group sparing works per disk group and only protects Vraid 1 and Vraid 5 LUNs.

A Disk Group Sparing Value of 0 - means no protection - although vraid 1 & vraid 5 luns can still be protected against members failure if there is sufficient unallocated disk space available in the disk group. The way the sparing algoritm on EVAs work - it always first looks at the unallocated disk space in the disk group - the last resort is the reserved space (as dictated by the sparing value per disk group). So if you have plenty of unallocated disk space in your disk group - you might survive Vraid 1 & Vraid 5 - disk members failure before the EVA even uses the reserved space for disk mechanism failure protection.

Any Vraid 0 LUNs will be lost regardless of the sparing parameter + unallocated free space in the disk group.

A sparing value of 1 reserves 2 times the size of the largest disk in your disk group - and similarly 4 times the size of the largest disk in your disk group for sparing value 2.
Reason for the double multiplier value - comes from the fact that Vraid 1 is implemented in "married pairs" - so even a single disk failure for a Vraid 1 Vdisk warrants movement of data for both the disk members of the married pair and hence disk space equal to 2*the size of the largest disk needs to be reserved.

Hope it is of help.