[Gllug] RAID 1+0 vs 0+1

[Rich Walker]

Richard Huxton <dev at archonet.com> writes:

>
> Hmm, I'd expect better performance on one disk failure with 1+0
> too. You've got 5 spindles still running versus 3 for 0+1, and a 2/3
> chance of reading from a mirrored disk.

AIUI, most significant array failures happen not when one disk goes
while the system is running; they happen when you shut down a server
that's been on 24x7 for the last year or so, and you get *multiple*
failures on power-up.

There's probably some neat statistical analysis that can be done to
prove the point, but:

(1|1) + (1|1) + (1|1) : for a 2-disk failure, you need both disks to be
in the same pair to bring down the array.

(1+1+1) | (1+1+1) : for a 2-disk failure, you need both disks to be in
the same triad to *not* bring down the array.

Assuming that the failures are independent but happen one after the
other, in the first case you have a 1 in 5 chance of catastrophe, in the
second case you have a 3 in 5 chance of catastrophe.

> Of course, the point is not to sit there admiring your good
> performance, it's to run and fetch another disk.

Here's hoping none of the others go...

cheers, Rich.

>
> -- 
>    Richard Huxton
>    Archonet Ltd
> -- 
> Gllug mailing list  -  Gllug at gllug.org.uk
> http://lists.gllug.org.uk/mailman/listinfo/gllug

-- 
rich walker         |  Shadow Robot Company | rw at shadow.org.uk
technical director     251 Liverpool Road   |
need a Hand?           London  N1 1LX       | +UK 20 7700 2487
www.shadow.org.uk/products/newhand.shtml
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug