[volunteergrid2-l] Slashdot post

[Eugen Leitl]

On Thu, Apr 05, 2012 at 06:08:14AM -0600, Shawn Willden wrote:

> One comment here:  I've come to the conclusion that RAID5 is evil and
> dangerous.  I've been meaning to write a paper on it for years but never
> gotten around to it.  Assuming you back everything important up to the grid
> it would be okay, but personally I use RAID6 for everything important.

The failure mode in large, consumer (high unrecoverable error rate)
unscrubbed (undetected bit rot) is that one drive falls from the array,
and during rebuild/resilver either due to bit rot or uncaught error
the second drive falls from the array, after which your data is typically
toast.

Higher RAID levels (RAID10, RAID6, raidz2/raidz3) will protect you
from that. I would still recommend weekly scrubbing for consumer
(SATA is consumer) drives.

I recommend using zfs and raidz2 or higher in suitably designed 
arrays with weekly or at least monthly (for SAS) scrubbing for 
more or less important data. Notice that with zfs send you can
easily replicate a consistent snapshot to a local or remote zfs
pool, with a high peformance than rsync or unionfs.

A good cheap hardware platform for that is HP N40L with 8 GByte ECC RAM
and a Solaris derviate with napp-it. Same platform will also support
FreeNAS, a FreeBSD distro supporting zfs (but less than Solaris 
derivates).