As one of the people recommending early investigation of Intel's recent
320 series drives, I've been following the news around them too. It
looks like there's one serious firmware bug that shows up on these so
far, what's being called the "8MB bug". Basically, under some
conditions, the drive comes back from a restart believing it's only 8MB
in size. Very bad.
In the discussion forum where this been highlighted:
http://communities.intel.com/message/131328 the largest data point I
noticed said that a large deployment has had 7 out of their 600 320
drive deployments go bad in this way, so 1.2%. Another report says 13
out of their 64 systems are dead now. You did never even consider
deploying this drive unless it was with RAID-1 and good real-time
backups, right?
Hopefully everyone knows by now that the V1 of any hardware should sit
in QA for a while to shake out issues like this before you move
production onto it, and this one seems to be the big bug in this
design. It seems like a firmware bug that an update will fix, not a
more serious design issue, so I'm not too concerned about it yet.
http://communities.intel.com/thread/23217 is where the official
commentary on the resolution is being posted to.
P.S. they have increased the warranty on these drives to 5 years, before
this all happened, so Intel has made a large bet on this model working
as advertised:
http://newsroom.intel.com/community/intel_newsroom/blog/2011/05/19/chip-shot-new-5-year-limited-warranty-on-intel-ssd-320
We just need to figure out how/where they're drawing the "enterprise
usage levels" line at.
--
Greg Smith 2ndQuadrant US greg@xxxxxxxxxxxxxxx Baltimore, MD
--
Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
