topic Re: EVA 4100 in Disk Enclosures

EVA 4100

AIG — Wed, 15 Sep 2010 09:48:54 GMT

Hi

I have a EVA4100 2C3D. I have a doubt about the EVA. we have all the disks configured in the default disk group.we have vdisks with raid 1 and raid 5. how are the vdisks assigned on the physical disks and how many disk failure on the EVA will cause data loss.

Thanks in advance

Re: EVA 4100

Oscar_Maimó — Wed, 15 Sep 2010 11:56:42 GMT

It depends of your RSS map. All vdisk redundancy is contained within a RSS set. If you lose one disk you will lose all VRAID 0 vdisks. If you lose two disks within a RSS you will lose your VRAID 5 data. If the disks are in pair you will lose your VRAID 1 data too.

Greetings
Oscar
(http://oscarmaimo.blogspot.com)

Re: EVA 4100

AnthonySN — Wed, 15 Sep 2010 12:30:17 GMT

what is the disk protection level(single or double) you have set on the disk group?

Re: EVA 4100

AIG — Wed, 15 Sep 2010 14:58:05 GMT

we have set it to double

Re: EVA 4100

Rob Leadbeater — Wed, 15 Sep 2010 19:48:04 GMT

Hi,

The disk protection level doesn't give you any more redundancy. It simply reserves more space within the group which is used in case disks fail.

In an EVA data for all vdisks is distributed across all physical disks in the group.

Assuming all the shelves in your EVA are populated, you'll have 42 disks which will probably be split into 5 RSS groups. The EVA tries to keep RSS groups to an optimum size of 8 disks, so you'll possibly have 4 RSS groups with 8 disks and 1 of 10 disks - the exact layout may be different depending on how disks have failed in the past.

As mentioned in a previous reply, if two disks within the same RSS group fail, then you'll lose any Vraid-5 vdisks. If two disks that are paired within the same RSS group fail, then you'll lose any Vraid-1 vdisks.

So in the worst case, two disks fail and you could lose everything.

In the best case, 5 disks (one from each RSS group) could fail and everything would carry out on working fine.

That is why the best practice is always to replace disks as soon as you can after they fail, because no-one can predict which disk is going to fail next...

Regards,

Rob

Re: EVA 4100

Oscar_Maimó — Thu, 16 Sep 2010 12:09:22 GMT

Rob:
When you said:
"In the best case, 5 disks (one from each RSS group) could fail and everything would carry out on working fine."
You must take into account the free space needed in Disk Group for reconstruction. In this case the spare and free space comes to play.
Greetings
Oscar