Is anyone here running a MAID system? Copan is the first one that
comes to mind.

I'm interested in experience with a dense HDD system that powers off
and powers on the drives for use and occasional checking.

In theory and concept this is a good thing, particularly with archive
or even daily backups.

Thanks.

~F

I've never been able to understand how they can guarantee that a disk
would power down and power back up with no problems whatsoever.

Call me skeptical.


On May 8, 2:56 pm, Faeandar <mr_casta... (AT) yahoo (DOT) com> wrote:

Sandman wrote:
I don't think they would ever make such claim, any more than any array
vendor would. Disks die, the trick is to make sure that you don't lose
data when they die, and that's not that hard to achieve these days. So
yes, they probably have a higher mortality rate for drives because of
the nature of what they do, but the drives are cheap and the power
savings substantial.



--
Nik Simpson

On 15 May 2007 00:06:14 -0700, Sandman <enjoylife_95135 (AT) hotmail (DOT) com>
wrote:


Well, I power my laptop and desktop off and on several times a day.
My laptop has bought the farm once and my desktop not at all. So I
would say that some classes of drives are ok with this type of
activity.
And as Nik stated, they are not making the claim you are skeptical
about. Just as HDS or NetApp or EMC will never claim none of their
drives will fail, neither do MAID vendors.

Given the market these vendors are targetting I see it as a viable
option. Extreme density, low power and cooling requirements, and
nearline response times.

I'm guessing from the lack of responses to the original query that no
one is actually using a MAID, yet.

~F

Sandman wrote:
As already noted by others, they don't. However, the recent FAST papers
indicate that rather than having a conventional bathtub failure graph
during their specified lifetimes disks fail with greater frequency later
in their service life, so keeping them powered down for a large
percentage of the time should reduce such wear-related failures and/or
allow service life to be extended beyond the nominal 5 years.

Furthermore, since many disks are rated for 50K start/stop cycles over
their lifetime, they could be restarted every hour without exceeding
their nominal specs. I wouldn't be surprised to see *some* increase in
related failures if that were done, but starting them once every day or
three should be a cakewalk (my impression is that Copan fires theirs up
perhaps about once every week or three to 'exercise' them if other
activity hasn't).

Couple the above with the significant power savings and you've got a
worthwhile product for people who don't need instantaneous access to
their data (disk backups being a prime example, of course). I expect
the technology to migrate into on-line storage eventually (e.g., with
mirrored storage and intelligent delayed write-back nearly half your
disks can be powered down virtually all the time, as long as you don't
need mirror copies for throughput enhancement).

- bill

On May 15, 7:59 pm, Bill Todd <billt... (AT) metrocast (DOT) net> wrote:

I don't know for sure, but I always thought they were powering the
drives down to a sleep state, not power off. Powering up from there
may well yield substantially less of a "power shock" that a real power
on, particularly to the electronics.

On May 16, 12:56 am, "robertwess... (AT) yahoo (DOT) com"
<robertwess... (AT) yahoo (DOT) com> wrote:

there is a white Paper on Copans website (www.copansys.com) that shows
that the powering off of drives actually drastically increases there
mean time between failure (MTBF) numbers for the drive. This has also
been confirmed by Seagate, Copans drive maker. Exercising the drives,
or Disk Arobics as copan calls it, will exercise each drive about once
a month and detects defects and repairs any bad blocks it finds. We
currently use a 4 shelf copan VTL system with 500GB drives which gives
us 162TB (usable) in 16U of space. its been pretty reliable and we
plan to upgrade to more space as our data needs grow. You just can't
beat the density. I'd be happy to answer any questions about the
system that I can

AJ

Just to make it clear, my intention was not to ridicule or deny
Copan's claims in any way. But I appreciate the discussion and
comments!

On May 15, 1:13 pm, Faeandar <mr_casta... (AT) yahoo (DOT) com> wrote: