RAID 1 or Hammer

[Dmitri Nikulin]

On Tue, Jan 13, 2009 at 9:20 AM, Simon 'corecode' Schubert
<corecode at fs.ei.tum.de> wrote:
> You're assuming fail-stop errors.  If it is a sneaking bit error or
> something else, it won't notice.

Ever since ZFS was announced I've wondered how often this actually
happens. So far I've only heard of one anecdote of silent readable
corruption which was caused by a faulty power supply.

Never in all my years of cheap disks and cheaper motherboards and
power supplies have I had a checksum on a file go bad, and I store
checksums on almost everything.

I did once have some live ext3 corruption but come on, it's ext3, and
that too could be explained by my power supply which later took out
two motherboards before its ritual suicide. Even then fsck.ext3
returned the fs to consistency and told me exactly what was damaged.

I'm just curious how much the average developer / admin would benefit
from ZFS. It seems it's good for really long term storage, since it'll
heal over bad blocks if scrubbed regularly. But how often do silent
bad blocks even occur?

-- 
Dmitri Nikulin

Centre for Synchrotron Science
Monash University
Victoria 3800, Australia