[57north-discuss] Harbinger of doom?

[Tony Travis]

Hi,

Rob was explaining to me why hard disks can't be trusted last night and,
although I disagreed with him about it, I do think he has a point and it
got me thinking a bit more deeply about the issue...

One thing we disagreed about was the extent to which the manufacturer's
UER (Unrecoverable Error Rate) spec. of a SATA drive can be used to
estimate how often you are likely to lose data from it. I found this
paper about the issue:

  Empirical Measurements of Disk Failure Rates and Error Rates
  http://arxiv.org/abs/cs/0701166

It's quite old now (2005) but I found it an interesting read:

Abstract:

The SATA advertised bit error rate of one error in 10 terabytes is
frightening. We moved 2 PB through low-cost hardware and saw five disk
read error events, several controller failures, and many system reboots
caused by security patches. We conclude that SATA uncorrectable read
errors are not yet a dominant system-fault source – they happen, but are
rare compared to other problems. We also conclude that UER
(uncorrectable error rate) is not the relevant metric for our needs.
When an uncorrectable read error happens, there are typically several
damaged storage blocks (and many uncorrectable read errors.) Also, some
uncorrectable read errors may be masked by the operating system. The
more meaningful metric for data architects is Mean Time To Data Loss
(MTTDL.)

I'd like to get a realistic idea of how worried we should be about Rob's
pronouncements and I wonder if anyone else in the Hackspace might be
interested in doing some empirical experiments to support or refute our
Harbinger of doom?

Thanks for an interesting discussion Rob - I was listening :-)

  Tony.

-- 
Minke Informatics Limited, Registered in Scotland - Company No. SC419028
Registered Office: 3 Donview, Bridge of Alford AB33 8QJ, Scotland (UK)
tel. +44(0)19755 63548                    http://minke-informatics.co.uk
mob. +44(0)7985 078324        mailto:tony.travis at minke-informatics.co.uk