Re: New system recommendations - Mailing list pgsql-admin

From Scott Marlowe
Subject Re: New system recommendations
Date
Msg-id 1146164727.23538.322.camel@state.g2switchworks.com
Whole thread Raw
In response to Re: New system recommendations  (Robin Iddon <robin@edesix.com>)
List pgsql-admin
On Thu, 2006-04-27 at 13:43, Robin Iddon wrote:
> > Anytime you're looking at a high write load on a database (not just
> > PostgreSQL) you should be looking at a hardware RAID controller with
> > battery backed cache and RAID 1 or RAID 1+0.
> >
> We recently benchmarked the latest and greatest PCI-X SATA mega raid
> controller (with NCQ support, so approaching SCSI performance) and we
> found it was no faster than running s/w RAID-1 on a dual-core P4 with
> the 2.6 series kernel.  It would no doubt offer an advantage on running
> RAID-5 or any other RAID scheme that required XOR computation, but we
> don't want to use RAID-5.
>
> However we did not fit the Transportable Battery Backup Unit and so we
> didn't measure the benefit of early commit to the cache (fsync() returns
> early, data is written to disk later, battery keeps data safe in the
> meanwhile).

That battery backed cache makes all the difference.  Also, unless
someone's done some work on it, the older linux kernel raid modules
serialized the access via multi-level RAID in such a way that RAID 1+0
was no faster than RAID 1.  You're right about RAID5 sw versus hw.
Without the BBCache unit, they're pretty much the same, and generally
limited by the speed of the drives, not the method of implementation.

Note that the Areca controllers and a few others get much better
scores.  But, the megaraid and adaptec controllers are the only ones
dell includes in their boxen, so if it's a choice between the adaptec
and the megaraid, I'd normally pick the megaraid.  Note that some folks
have mentioned that Dell uses their own firmware on these cards, and
that firmware is supposedly slower than the stock firmware those cards
come with.

> I plan on repeating the experiment again later on in May once we have a
> TBBU and will post the results here if anyone is interested.

I'd love to see how the newer cards perform.  I know that for RAID 1,
straight up, linux kernel sw RAID is quite speedy.  I'd imagine it would
take a fairly heavily written environment to see an advantage for the HW
controller.

pgsql-admin by date:

Previous
From: Robin Iddon
Date:
Subject: Re: New system recommendations
Next
From: Chris Browne
Date:
Subject: Re: New system recommendations