Subject: Re: most bang for buck with ~ $20,000
From: Scott Marlowe <smarlowe@g2switchworks.com>
To: "Jim C. Nasby" <jnasby@pervasive.com>
Cc: "Thomas F. O'Connell" <tfo@sitening.com>,
	"Joshua D.Drake" <jd@commandprompt.com>,
	Kenji Morishige <kenjim@juniper.net>, pgsql-performance@postgresql.org
In-Reply-To: <20060809213543.GV40481@pervasive.com>
References: <20060808204354.GN29379@juniper.net>
	<44D906E1.7010405@commandprompt.com>
	<9EE66C7F-0B04-4472-9EBD-CD567AC9EA1D@sitening.com>
	<44D90FFE.7010807@commandprompt.com>
	<82C06632-148B-4489-90EB-BBB8A6BD1159@sitening.com>
	<1155136527.20252.90.camel@state.g2switchworks.com>
	<20060809213543.GV40481@pervasive.com>
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Message-Id: <1155160230.20252.124.camel@state.g2switchworks.com>
Mime-Version: 1.0
Date: Wed, 09 Aug 2006 16:50:30 -0500

On Wed, 2006-08-09 at 16:35, Jim C. Nasby wrote:
> On Wed, Aug 09, 2006 at 10:15:27AM -0500, Scott Marlowe wrote:
> > Actually, the BIGGEST win comes when you've got battery backed cache on
> > your RAID controller.  In fact, I'd spend money on a separate RAID
> > controller for xlog with its own cache hitting a simple mirror set
> > before I'd spring for more drives on pg_xlog.  The battery backed cache
> > on the pg_xlog likely wouldn't need to be big, just there and set to
> > write-back.
> > 
> > Then put all the rest of your cash into disks on a big RAID 10 config,
> > and as big of a battery backed cache as you can afford for it and memory
> > for the machine.
> 
> Actually, my (limited) testing has show than on a good battery-backed
> controller, there's no penalty to leaving pg_xlog in with the rest of
> PGDATA. This means that the OP could pile all 8 drives into a RAID10,
> which would almost certainly do better than 6+2.

I've seen a few posts that said that before.  I wonder if there's a
point where the single RAID array / controller would get saturated and a
second one would help.  I think most of the testing I've seen so far has
been multiple RAID arrays under the same controller, hasn't it?

> Note that some controllers (such as 3ware) need to periodically test the
> life of the BBU, and they disable write caching when they do so, which
> would tank performance.

ugh, that's a scary thing.  Can you at least schedule it?