[zfs-discuss] zfs-discuss Digest, Vol 43, Issue 3

Gordan Bobic gordan.bobic at gmail.com
Wed Nov 7 17:56:57 EST 2018


For obvious reasons it is not possible for any current model to have ever
had millions of hours of testing per drive. :-)

Backblaze are publishing their stats specifically because they are too
small to buy direct from the manufacturer (10,000 drives at a time). Bigger
players like Google have to anonymize their findings by makes/models
specifically because they would lose their discounts. It is most fortunate
for all of us that Backblaze are publishing their reliability stats.

On Wed, 7 Nov 2018, 21:55 Bzzzz via zfs-discuss <
zfs-discuss at list.zfsonlinux.org wrote:

> On Wed, 07 Nov 2018 22:03:49 +0100
> Gionatan Danti <g.danti at assyoma.it> wrote:
>
> > Interesting... and condivisible. [my italian is as rusty as my german]
>
> > Surely if the manufacturers lie (and
> > often they do!) we have a problem.
> > Anyway, give a look here:
> > https://www.backblaze.com/blog/smart-22-is-a-gas-gas-gas/
>
> Hum, one part of the 2nd sentence is pulling attention:
> "allows the drives to run cooler and quieter"
> cooler, yes, as He has a very low resistance to temperature,
> but quieter, I doubt (which pose a question: what is the real
> relationship between backblaze and WD ? Well, to be more precise,
> another one too: do you really think such a big customer couldn't
> get an extra few percent off praising he's supplier's product ?)
>
> My contention is to wait a bit more until He HDz other large consumers
> check or not what backblaze says.
>
> Something makes me uncomfortable in the commercial brochure, though:
>
> https://www.hgst.com/sites/default/files/resources/HGST-Helium-Technology-BR03.pdf
> P.3, §2: "Prior to shipping helium-filled drives, we performed long-
> term reliability testing, with 5,000 drives each running more than 1,000
> hours (over 5 million combined hours), showing no seal degradation on
> any of the drives."
>
> It is not said, but suggested, in the right line of neuro-science
> studies (mostly paid by the marketing industry to try to learn how to
> fool people an easier way) - the incriminated part is the one about
> the "5M combined hours".
>
> Just as if I told you to buy my new car because I made tests using
> 20,000 of them, making 1000 km each…
> (20,000,000 combined kilometers ;-p)
>
> What is also concerning is the SMART indicator set into the pre-fail
> type, meaning when your He HD rings the bell, you're doomed because the
> drive aerodynamics was calculated for Helium fluid, not air.
>
> May be vacuum instead of Helium would have been a better idea
> (disclaimer: I'm not a fluids mechanic specialist.)
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at list.zfsonlinux.org
> http://list.zfsonlinux.org/cgi-bin/mailman/listinfo/zfs-discuss
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.zfsonlinux.org/pipermail/zfs-discuss/attachments/20181107/a0d5c1b1/attachment.html>


More information about the zfs-discuss mailing list