pg.ddx.io pgsql-performance@postgresql.org mailing list archive
help / color / mirror / Atom feedBattery Backed Cache for RAID
8+ messages / 8 participants
[nested] [flat]
* Battery Backed Cache for RAID
@ 2005-09-14 18:25 Peter Darley <pdarley@kinesis-cem.com>
0 siblings, 1 reply; 8+ messages in thread
From: Peter Darley @ 2005-09-14 18:25 UTC (permalink / raw)
To: pgsql-performance
Folks,
I'm getting a new server for our database, and I have a quick question
about RAID controllers with a battery backed cache. I understand that the
cache will allow the cache to be written out if the power fails to the box,
which allows it to report a write as committed safely when it's not actually
committed.
My question is, if the power goes off, and the drives stop, how does the
battery backed cache save things out to the dead drives? Is there another
component that is implied that will provide power to the drives that I
should be looking into as well?
Thanks,
Peter Darley
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 18:28 Jeffrey W. Baker <jwbaker@acm.org>
parent: Peter Darley <pdarley@kinesis-cem.com>
0 siblings, 1 reply; 8+ messages in thread
From: Jeffrey W. Baker @ 2005-09-14 18:28 UTC (permalink / raw)
To: Peter Darley <pdarley@kinesis-cem.com>; +Cc: pgsql-performance
On Wed, 2005-09-14 at 11:25 -0700, Peter Darley wrote:
> I'm getting a new server for our database, and I have a quick question
> about RAID controllers with a battery backed cache. I understand that the
> cache will allow the cache to be written out if the power fails to the box,
> which allows it to report a write as committed safely when it's not actually
> committed.
Actually the cache will just hold its contents while the power is out.
When the power is restored, the RAID controller will complete the writes
to disk. If the battery does not last through the outage, the data is
lost.
> My question is, if the power goes off, and the drives stop, how does the
> battery backed cache save things out to the dead drives? Is there another
> component that is implied that will provide power to the drives that I
> should be looking into as well?
A UPS would allow you to do an orderly shutdown and write contents to
disk during a power failure. However a UPS can be an extra point of
failure.
-jwb
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 20:03 Alvaro Herrera <alvherre@alvh.no-ip.org>
parent: Jeffrey W. Baker <jwbaker@acm.org>
0 siblings, 2 replies; 8+ messages in thread
From: Alvaro Herrera @ 2005-09-14 20:03 UTC (permalink / raw)
To: Jeffrey W. Baker <jwbaker@acm.org>; +Cc: Peter Darley <pdarley@kinesis-cem.com>; pgsql-performance
On Wed, Sep 14, 2005 at 11:28:43AM -0700, Jeffrey W. Baker wrote:
> On Wed, 2005-09-14 at 11:25 -0700, Peter Darley wrote:
> > I'm getting a new server for our database, and I have a quick question
> > about RAID controllers with a battery backed cache. I understand that the
> > cache will allow the cache to be written out if the power fails to the box,
> > which allows it to report a write as committed safely when it's not actually
> > committed.
>
> Actually the cache will just hold its contents while the power is out.
> When the power is restored, the RAID controller will complete the writes
> to disk. If the battery does not last through the outage, the data is
> lost.
Just curious: how long are the batteries supposed to last?
--
Alvaro Herrera -- Valdivia, Chile Architect, www.EnterpriseDB.com
Hi! I'm a .signature virus!
cp me into your .signature file to help me spread!
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 20:17 Merlin Moncure <merlin.moncure@rcsonline.com>
0 siblings, 0 replies; 8+ messages in thread
From: Merlin Moncure @ 2005-09-14 20:17 UTC (permalink / raw)
To: Alvaro Herrera <alvherre@alvh.no-ip.org>; +Cc: pgsql-performance
> On Wed, Sep 14, 2005 at 11:28:43AM -0700, Jeffrey W. Baker wrote:
> > On Wed, 2005-09-14 at 11:25 -0700, Peter Darley wrote:
> > > I'm getting a new server for our database, and I have a quick
> question
> > > about RAID controllers with a battery backed cache. I understand
that
> the
> > > cache will allow the cache to be written out if the power fails to
the
> box,
> > > which allows it to report a write as committed safely when it's
not
> actually
> > > committed.
> >
> > Actually the cache will just hold its contents while the power is
out.
> > When the power is restored, the RAID controller will complete the
writes
> > to disk. If the battery does not last through the outage, the data
is
> > lost.
>
> Just curious: how long are the batteries supposed to last?
For the length of time it will take for you to get fired for not getting
the server running plus one hour :).
Merlin
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 20:40 John A Meinel <john@arbash-meinel.com>
parent: Alvaro Herrera <alvherre@alvh.no-ip.org>
1 sibling, 0 replies; 8+ messages in thread
From: John A Meinel @ 2005-09-14 20:40 UTC (permalink / raw)
To: Alvaro Herrera <alvherre@alvh.no-ip.org>; +Cc: Jeffrey W. Baker <jwbaker@acm.org>; Peter Darley <pdarley@kinesis-cem.com>; pgsql-performance
Alvaro Herrera wrote:
> On Wed, Sep 14, 2005 at 11:28:43AM -0700, Jeffrey W. Baker wrote:
>
>>On Wed, 2005-09-14 at 11:25 -0700, Peter Darley wrote:
>>
>>> I'm getting a new server for our database, and I have a quick question
>>>about RAID controllers with a battery backed cache. I understand that the
>>>cache will allow the cache to be written out if the power fails to the box,
>>>which allows it to report a write as committed safely when it's not actually
>>>committed.
>>
>>Actually the cache will just hold its contents while the power is out.
>>When the power is restored, the RAID controller will complete the writes
>>to disk. If the battery does not last through the outage, the data is
>>lost.
>
>
> Just curious: how long are the batteries supposed to last?
>
The recent *cheap* version of a ramdisk had battery backup for 16 hours.
(Very expensive ramdisks actually have enough battery power to power a
small hard-drive to dump the contents into).
I'm guessing for a RAID controller, the time would be in the max 1 day
range.
John
=:->
Attachments:
signature.asc (application/pgp-signature, 252B) download
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 20:46 Welty, Richard <richard.welty@bankofamerica.com>
0 siblings, 0 replies; 8+ messages in thread
From: Welty, Richard @ 2005-09-14 20:46 UTC (permalink / raw)
To: John A Meinel <john@arbash-meinel.com>; +Cc: pgsql-performance
John A Meinel wrote:
>The recent *cheap* version of a ramdisk had battery backup for 16 hours.
>(Very expensive ramdisks actually have enough battery power to power a
>small hard-drive to dump the contents into).
>I'm guessing for a RAID controller, the time would be in the max 1 day
>range.
i think some will go a bit longer. i have seen an IBM ServeRaid (rebranded
mylex in this particular case) keep its memory after being pulled for a
remarkably long period of time.
no guarantees, though, so i'm not actually going to say how long so that nobody
gets unreasonable expectations.
richard
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 20:47 Arjen van der Meijden <acmmailing@tweakers.net>
parent: Alvaro Herrera <alvherre@alvh.no-ip.org>
1 sibling, 1 reply; 8+ messages in thread
From: Arjen van der Meijden @ 2005-09-14 20:47 UTC (permalink / raw)
To: Alvaro Herrera <alvherre@alvh.no-ip.org>; +Cc: Jeffrey W. Baker <jwbaker@acm.org>; Peter Darley <pdarley@kinesis-cem.com>; pgsql-performance
On 14-9-2005 22:03, Alvaro Herrera wrote:
> On Wed, Sep 14, 2005 at 11:28:43AM -0700, Jeffrey W. Baker wrote:
>
>>On Wed, 2005-09-14 at 11:25 -0700, Peter Darley wrote:
>>
>>Actually the cache will just hold its contents while the power is out.
>>When the power is restored, the RAID controller will complete the writes
>>to disk. If the battery does not last through the outage, the data is
>>lost.
>
>
> Just curious: how long are the batteries supposed to last?
For the LSI-Logic MegaRaid 320-2e its about 72 hours for the standard
128MB version. Their SATA2-solution offers 32 and 72 hour-options. So I
assume its "in the order of days" for most RAID controllers.
Best regards,
Arjen van der Meijden
^ permalink raw reply [nested|flat] 8+ messages in thread
* Re: Battery Backed Cache for RAID
@ 2005-09-14 21:58 mudfoot@rawbw.com
parent: Arjen van der Meijden <acmmailing@tweakers.net>
0 siblings, 0 replies; 8+ messages in thread
From: mudfoot@rawbw.com @ 2005-09-14 21:58 UTC (permalink / raw)
To: pgsql-performance
Bear in mind you will lose data if the raid controller itself fails (or the
cache memory module). Many solutions have mirrored cache for this reason. But
that's more $$, depending on the risks you want to take.
Quoting Arjen van der Meijden <acmmailing@tweakers.net>:
> On 14-9-2005 22:03, Alvaro Herrera wrote:
> > On Wed, Sep 14, 2005 at 11:28:43AM -0700, Jeffrey W. Baker wrote:
> >
> >>On Wed, 2005-09-14 at 11:25 -0700, Peter Darley wrote:
> >>
> >>Actually the cache will just hold its contents while the power is out.
> >>When the power is restored, the RAID controller will complete the writes
> >>to disk. If the battery does not last through the outage, the data is
> >>lost.
> >
> >
> > Just curious: how long are the batteries supposed to last?
>
> For the LSI-Logic MegaRaid 320-2e its about 72 hours for the standard
> 128MB version. Their SATA2-solution offers 32 and 72 hour-options. So I
> assume its "in the order of days" for most RAID controllers.
>
> Best regards,
>
> Arjen van der Meijden
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: explain analyze is your friend
>
^ permalink raw reply [nested|flat] 8+ messages in thread
end of thread, other threads:[~2005-09-14 21:58 UTC | newest]
Thread overview: 8+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2005-09-14 18:25 Battery Backed Cache for RAID Peter Darley <pdarley@kinesis-cem.com>
2005-09-14 18:28 ` Jeffrey W. Baker <jwbaker@acm.org>
2005-09-14 20:03 ` Alvaro Herrera <alvherre@alvh.no-ip.org>
2005-09-14 20:40 ` John A Meinel <john@arbash-meinel.com>
2005-09-14 20:47 ` Arjen van der Meijden <acmmailing@tweakers.net>
2005-09-14 21:58 ` mudfoot@rawbw.com
2005-09-14 20:17 Re: Battery Backed Cache for RAID Merlin Moncure <merlin.moncure@rcsonline.com>
2005-09-14 20:46 Re: Battery Backed Cache for RAID Welty, Richard <richard.welty@bankofamerica.com>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox