Is it a good idea to use 2.5" drives in a server that's up 24x7?

One part of me says "less power, less heat, that's good for them and
the server and they are quieter too" and another part of me thinks
that notebook drives fail more often. Then again, they get carted
around all over the place, so their life in a notebook is kind of
rough compared to sitting a metal box in a corner their entire life.

Opinions? Reliability is my main goal. If 3.5's are more reliable, I
will go that way. POinters to any "life span" test sites appreciated.

Thanks,

me wrote:

> Is it a good idea to use 2.5" drives in a server that's up 24x7?

No evidence that they dont survive fine in that config.

> One part of me says "less power, less heat, that's good for them and the server

Thats easily overstated with servers with adequate cooling.

> and they are quieter too" and another part of me thinks that
> notebook drives fail more often. Then again, they get carted
> around all over the place, so their life in a notebook is kind of
> rough compared to sitting a metal box in a corner their entire life.

Precisely.

> Opinions? Reliability is my main goal. If 3.5's are more reliable, I will go that way.

They do appear to be.

> POinters to any "life span" test sites appreciated.

No one bothers to do that, essentially because by the
time the test is complete, the drives are obsolete anyway.

You do see some operations like google do some analysis of life spans seen, but
they dont generally use 2.5" drives, so thats not any practical use with this choice.

The other obvious downsides with 2.5" drives is that they are much more expensive
per GB, slower than 3.5" drives and arent available in the same largest sizes either.

me <(E-Mail Removed)> kenjka:
> Is it a good idea to use 2.5" drives in a server that's up 24x7?

HP DL3xx series use 2.5" drives and have no 3.5" bays... And it's the most
popular HP server series...

If you need reliability, forget using single-drive configurations, go with
RAID array of minimum 2 drives in mirror (RAID1) or 3 drives in parity
(RAID5)... Also, consider using hotswappable drive bays...

--
U autobusu se cijeli dan optuzen balijaog prdija. By runf

Damir Lukic, calypso@_MAKNIOVO_fly.srk.fer.hr
http://inovator.blog.hr
http://calypso-innovations.blogspot.com/

me <(E-Mail Removed)> wrote:
> Is it a good idea to use 2.5" drives in a server that's up 24x7?

Depends.

> One part of me says "less power, less heat, that's good for them and
> the server and they are quieter too" and another part of me thinks
> that notebook drives fail more often. Then again, they get carted
> around all over the place, so their life in a notebook is kind of
> rough compared to sitting a metal box in a corner their entire life.

> Opinions? Reliability is my main goal. If 3.5's are more reliable, I
> will go that way. POinters to any "life span" test sites appreciated.

I don't think that under the same conditions 3.5" is more reliable.
3.5" is faster though at the same rotational speed. 3.5" is also
cheaper.

That said, I have had a fileserver/firewall (old PC with Linux) with
2.5" notebook drives, that has a 3-way RAID1 setup for important
backups for something like 7 years now. The heat generated is much
less. Noise is lower, however mechanical decoupling can still
be needed. The only problem I have had so far was a drive with (rare)
SATA disconnects, nothing I can identidy as a 2.5" problem.

I think that notebook drives are tested for endurance under different
conditions that 3.5" drives. Far more heat for one thing. Mechanical
shock for another. That would explain the shorter warranty times
adequately.

Arno

--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: (E-Mail Removed)
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans

On Tue, 27 Apr 2010 22:26:38 -0400, me <(E-Mail Removed)> put
finger to keyboard and composed:

>Is it a good idea to use 2.5" drives in a server that's up 24x7?
>
>One part of me says "less power, less heat, that's good for them and
>the server and they are quieter too" and another part of me thinks
>that notebook drives fail more often. Then again, they get carted
>around all over the place, so their life in a notebook is kind of
>rough compared to sitting a metal box in a corner their entire life.
>
>Opinions? Reliability is my main goal. If 3.5's are more reliable, I
>will go that way. POinters to any "life span" test sites appreciated.
>
>Thanks,

Notebook drives would have far more aggressive APM settings. This
means that they could park their heads several times per minute, which
in turn means they could very quickly exceed their rated load/unload
cycles, especially if operated at 24/7.

AIUI, some operating systems exacerbate this problem. Linux is
particularly badly behaved in this regard, although there are
workarounds.

See the large counts in this thread:
http://sourceforge.net/mailarchive/f...ntools-support

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

Franc Zabkar <(E-Mail Removed)> wrote:
> On Tue, 27 Apr 2010 22:26:38 -0400, me <(E-Mail Removed)> put
> finger to keyboard and composed:

>>Is it a good idea to use 2.5" drives in a server that's up 24x7?
>>
>>One part of me says "less power, less heat, that's good for them and
>>the server and they are quieter too" and another part of me thinks
>>that notebook drives fail more often. Then again, they get carted
>>around all over the place, so their life in a notebook is kind of
>>rough compared to sitting a metal box in a corner their entire life.
>>
>>Opinions? Reliability is my main goal. If 3.5's are more reliable, I
>>will go that way. POinters to any "life span" test sites appreciated.
>>
>>Thanks,

> Notebook drives would have far more aggressive APM settings. This
> means that they could park their heads several times per minute, which
> in turn means they could very quickly exceed their rated load/unload
> cycles, especially if operated at 24/7.

> AIUI, some operating systems exacerbate this problem. Linux is
> particularly badly behaved in this regard, although there are
> workarounds.

> See the large counts in this thread:
> http://sourceforge.net/mailarchive/f...ntools-support

Good point. I seem to have forgotten this and have
578576 load cycles on one drive after less than 2 years.

A "hdparm -B 254 /dev/<drive>" should fix this though.
....
Yes, works. No additional Load Cycles in 30 minutes
(before about 1 per 90 sec).

Arno
--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: (E-Mail Removed)
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans