Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

Protection against Disk failures on EVAs

k20pattu
Occasional Advisor

Protection against Disk failures on EVAs

Hello,

Another thread on a different topic and a few different questions. Could not find the definite answers from documents I've.

One diskgroup - 8 enclosures with 8 drives in each. RSSs look good - one disk from each shelf.
1. When I create a vdisk of raid0, will it be spread across all 64 drives? Does RSS play any role in deciding which disks will be used?
2. When I create a vdisk of raid1, will it be spead across 32 + 32 drives? Does RSS play any role in deciding which disks will be used?
3. Disk protection level set to double. Two drive from different enclosures fails within a span of 2 hours. RSS protects against the double failure from same enclousre.
If 3rd disk fails from 3rd enclosure, do we lose all data? Does RSS play any role here?
If 3rd disk fails from one of the previously failed drive enclosure, do we lose all data? Does RSS play any role here?

If you have a document that narrates about this, please send me the link.

Thanks

Ketan
2 REPLIES
Uwe Zessin
Honored Contributor

Re: Protection against Disk failures on EVAs

A Redundant Storage Set (RSS) is used to divide a disk group into multiple sub-groups. The redundancy information is stored in the same RSS where the data is located. The idea is:

- data and redundancy information is stored on a limited number of disks (between 6 and 11), so reconstruction can be done faster

- theoretically, a large disk group can sustain multiple disk failures -- as long as only one disk drive (or two when using VRAID-6) per RSS fails


BUT: RSS *DOS NOT* provide redundancy !!
Create a virtual disk in VRAID-0 and *ANY* failure of a disk drive causes data loss!!
The redundancy is ONLY provided by a VRAID-level other than 0.


1. Yes and Yes.

2. Yes and Yes. The data layout is a bit more complex, but as an end-user you do not really care and you can't incfluence it anyway.

3. The so-called "protection level" *DOES NOT* 'protect' any data! It is just the virtualized equivalent of spare disks.
As I wrote above - it depends on the VRAID-level and _which_ disk drives fail and if the previous reconstruct has finish whether you data is safe or not.


You disk group example with 8 disk drives has one RSS. Only when you go to 12 disk drives, it will be split into two RSS.
.
k20pattu
Occasional Advisor

Re: Protection against Disk failures on EVAs

Thats a good explanation, thanks for that. RSS distribution is visible to end users and only HP engineers can draw a diagram to show its layout and also it can't be changed. So basically, there is no way to say for sure, how many simultaneous drives an array can sustain before we lose the data?

The array is EVA8400 with 180 X 600FC.
Thanks
Ketan