[mythtv-users] BackBlaze hard drive study

Calvin Dodge caldodge at gmail.com
Thu Jan 30 13:40:44 UTC 2014


Mea culpa - Google DOES mention the significance of "infant mortality".

Calvin Dodge


On Thu, Jan 30, 2014 at 7:31 AM, Calvin Dodge <caldodge at gmail.com> wrote:

> "... high temperature is not as strongly correlated to failure as we
> typically expect."
>
> In addition, Google did not see an "infant mortality" spike with new
> drives.  Instead, they saw a linear failure rate over time for drives less
> than 3 years old (the failure rate increases after 3 years, due to bearing
> wear).
>
> The Backblaze article DOES mention that the average age of the 1.5 TB
> drives is 4 years (the "release date" tells you when a drive was first
> produced, not when a specific drive was built). To me it's more significant
> that BackBlaze's 4 TB Seagates average around 4 months in age, but have a
> greater annual failure rate than their 1-year-old Hitachis.
>
> I have one other semi-anecdotal bit of information to consider - my former
> employer has their services hosted at Tummy.com (a major Colorado hosting
> site), and Tummy swears by Hitachis. Since they have a fairly large number
> of hard drives in use, and they DO have properly-configured ventilation,
> etc., I grant them more credence than I do some consultant who probably
> does NOT have equivalent experience.
>
> Calvin Dodge
>
> Calvin Dodge"
>
>
> On Thu, Jan 30, 2014 at 12:06 AM, Karl Newman <newmank1 at asme.org> wrote:
>
>> On Thu, Jan 30, 2014 at 11:28 AM, Stephen P. Villano <
>> stephen.p.villano at gmail.com> wrote:
>>
>>>
>>> On 1/29/14, 9:23 PM, Gary Buhrmaster wrote:
>>> > On Tue, Jan 21, 2014 at 4:14 PM, Craig Treleaven <ctreleaven at cogeco.ca>
>>> wrote:
>>> >> Very interesting analysis from running 25,000 consumer-class drives
>>> over 5
>>> >> years:
>>> >>
>>> >> http://blog.backblaze.com/2014/01/21/what-hard-drive-should-i-buy/
>>> > A rather scathing response regard the statistical accuracy of this
>>> report:
>>> >
>>> http://www.enterprisestorageforum.com/storage-hardware/selecting-a-disk-drive-how-not-to-do-research-1.html
>>> > _______________________________________________
>>> >
>>> What I see is vendor data vs experiential data from one outfit that
>>> testing a ton of various inexpensive drives.
>>> As such, do we really want to trust an opinion that is based upon vendor
>>> ratings? Do we trust a tester who has no hard drive industry affiliation?
>>>
>>> Actually, neither.
>>> First, what was the HD temperature of each drive in each bundle in each
>>> unit during its testing?
>>> What was the ambient temperature of the rack, the array unit, etc?
>>>
>>> Those are really big deals, as we're talking about potentially cooking
>>> the hard drives, something unforgivable for any electronic device not
>>> designed to be cooked.
>>> And yes, I mean cooked. I've worked in far less than optimal
>>> environments, including server rooms that reached 105-110 degrees F and
>>> lousy ventilation, air conditioning performed by split air conditioner
>>> units designed for habitations, not a server room.
>>> We'll suffice it to say that we had significant loss of hardware,
>>> especially server fans and the occasional power supply.
>>> Add in the real difference between consumer hard drives and enterprise
>>> SAS drives, there's another really big difference.
>>> But, without temperature data, we really don't have a clue what was
>>> really going on.
>>> And to be blunt, I'll not trust my *own* word on it without a trainload
>>> of real numbers.
>>>
>>
>> Actually, Google's drive reliability study (
>> http://static.googleusercontent.com/media/research.google.com/en/us/archive/disk_failures.pdf)
>> indicated that high temperature is not as strongly correlated to failure as
>> we typically expect. Unfortunately (I'm sure lawyers intervened) Google
>> doesn't name names.
>>
>> Karl
>>
>> _______________________________________________
>> mythtv-users mailing list
>> mythtv-users at mythtv.org
>> http://www.mythtv.org/mailman/listinfo/mythtv-users
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mythtv.org/pipermail/mythtv-users/attachments/20140130/11a8f17e/attachment.html>


More information about the mythtv-users mailing list