Home made NAS

[Oleg Goldshmidt]

On Wed, Dec 5, 2012 at 9:46 AM, Moish <moish at mln.co.il> wrote:
> On 04/12/2012 21:27, E.S. Rosenberg wrote:
>>
>> According to research done by google and also in my experience a
>> normal harddisk (spinner) that has functioned without failures for 3
>> years will generally last for a very long time....
>
> Better read this article first (from 2007)
> http://storagemojo.com/2007/02/19/googles-disk-failure-experience/

As a word of warning, it is usually better to link to the actual paper
(quite a famous one in this case),
http://research.google.com/archive/disk_failures.pdf, and not to a
blog about it by someone who mixes up mean and median (he is honest
enough to acknowledge it after it was pointed out to him, but the text
was not changed).

To the point, there is very little that this particular paper says on
the topic of how much more likely old disks are to die. There is a
wealth of research papers on the subject, and the notion of "bathtube
curve" (new disks failing often - the so-called "infant mortality" -
then flat life expectancy and then old disks failing more often again)
is encountered commonly. If one assumes that the OPs "old" disks are
operational then it well may be that they have survived the "infancy"
and have some life in them.

One thing that the Google paper shows is that temperature
(overheating) affects older disks more than newer ones. One may assume
that a home setup in old boxes, with faulty fans, etc., this may
affect reliability adversely.

All of the above is probably negligible compared to two important
arguments that have already been mentioned: 1) as a home appliance
there are better, more economical, and - most importantly! - quieter
solutions for a modest price; 2) tinkering with such a heterogeneous
system will yield invaluable experience, especially in terms of never
trying anything like this for anything important.

-- 
Oleg Goldshmidt | pub at goldshmidt.org