EVA8400 VRAID5 redundancy

[Johnc_3]

Hi,

I've been asked the question about just what would happen with this array if several disks fail.  We currently have 98 disks in the VRAID5 disk group (with double protection) and have had two failures which occurred so quickly that they weren't replaced before the other failed.  Obviously, we've replaced them now and the group is currently leveling.  What would happen if another disk failed?

I know the 8400 arranges RSS by itself which would be 'robust' but the documentation doesn't explain just how robust this would be if further disks fail during the leveling.  Would data loss happen or would the move of data into sparing space help the data survive?

Would be interesting to find out a definitive answer.

Many thanks,

John

[Johan Guldmyr]

Hi!

Data loss would not happen because when the disk group is levelling the data has already been reconstructed or moved into the spare space (this would happen if the disk is not unreadable).

The important rebuild process is called reconstruction.  If a disk fails when levelling is going on that means reconstruction will happen and after that it will continue levelling.  Levelling levels out the data proportional to the disk size.

The double disk protection reserves free space, so that reconstruction goes faster.

With 98 disks in a VRAID5 you'll have many RSS groups, I haven't done the math now but this means if you're "lucky/unlucky" you could loose 9-10 disks without losing any data. With double spare you got plenty of reserved disk space to put the reconstructed data. leaving you vulnerable for a less period of time.