[LINK] Digital doomsday: the end of knowledge

[Andy Farkas]

On Wed, Apr 7, 2010 at 9:02 AM, Stilgherrian <stil at stilgherrian.com> wrote:
>
> I'd have thought that for hard drives the limiting factor wouldn't be
> the life of the magnetic medium but the mechanical life of the drive.
> Machines wear out. and "no-one knows"? I thought the mean time until
> failure for HDDs would be pretty well known by now -- and it'd be in
> a single digit of years. Can anyone confirm?
>

I recently lost a drive on my gateway box. I built the box back in 1999.
Its a Pentium-Pro 200MHz with 128MB RAM and had 3 x 2GB SCSI drives in a
software RAID5 setup. The box is still going strong, except it now has
2 x 9GB drives in a RAID1 configuration.

The drives had been spinning constantly for over 10 years, and every night
they would get a 'stir' when the OS scanned the file system, as well as
normal usage (writing and rotating logfiles, emails, squid). I couldn't
count the number of times I would wake at 3:01am and hear them rattling
away...

Anyway, the point is that it was a medium failure, not a mechanical one.


Some snippets from an interesting article I read recently:

http://blogs.zdnet.com/storage/?p=805&tag=rbxccnbzd1

"Why RAID 6 stops working in 2019"

"Three years ago I warned that RAID 5 would stop working in 2009. Sure
enough, no enterprise storage vendor now recommends RAID 5."

and

"You have a 62% chance of data loss due to an uncorrectable read error
on a 7 drive RAID with one failed disk, assuming a 10^14 read error rate
and ~23 billion sectors in 12 TB. Feeling lucky?"


Coincidentally, I've just bought 4 x 1.5TB WD15EADS drives and will be using
the amazing ZFS file system. I hope I get double-digit years out of them!
(but I doubt it)

-andyf