[lug] Hard Drive Manufacturer Suggestion

[Sean Reifschneider]

On Sun, Dec 05, 2004 at 11:49:25PM -0700, Ferdinand Schmid wrote:
>In the mid to late 90s we had many Western Digital drives (Caviar).  As others 
>already wrote we had many failures - eventually 100%.  I understand they had 
>gasket problems.

I've mostly avoided WD drives because of their weird "Master/Slave/Single"
settings that always seem to cause frustration.  They're the only drive
I've seen that requires being set in a special mode if they're the only
drive on a chain.  I don't think I've had any failures of the (very) few we
have though.

Most of the drives we've deployed in the last 5 years have been IBM and
IBM/Hitachi, and we've had very few problems.  That said, we did just have
an IBM drive totally fail that was exactly 3 years old (drive said
manufacture date of Nov 2001).  More commonly we've had the drives get into
a weird state where they'd act dead until they were power-cycled.  We've
been replacing those drives as we run across them, only maybe 5% of the
drives have had that problem so far.

This is out of something around 100 IBM drives, I'd guess.  Mostly PATA, a
few SCSI.  The SCSI drives that I still keep track of have been great.

The 1TB RAID SATA array I set up about a year ago was with 10 Seagate
drives.  The drives seem slow, but I haven't had any problems with them.
The 6-drive IBM PATA RAID-5 system benchmarks much faster than the 10-drive
Seagata SATA RAID-5.  Apparently, some of the early drives weren't very
good at performance.  Either one will out-perform the 100mbps I'm using at
it's fastest, so I don't really care, I was just comparing them.

Sean
-- 
 It's in that place where I put that thing that time.
                 -- The Phantom Phreak, _Hackers_
Sean Reifschneider, Member of Technical Staff <jafo at tummy.com>
tummy.com, ltd. - Linux Consulting since 1995.  Qmail, Python, SysAdmin