DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=message-id:date:from:to:subject:cc:in-reply-to:mime-version
	:content-type:content-transfer-encoding:content-disposition
	:references;
	b=EYI4H9RlfXpoHFvTYOFdP9j0O31udiKrtCnFhUcLLWotwGBucR1gPe6kW5UUEnlIJa
	9jHgEjxZpqVOWaq5Evkel+qc6UZdsAON/NWbVE4kaKh/dOhJefbCEh4iU4bc9eIL8HDI
	RxbdD3dhw0ick6FRsFlNlebVwwbuaTUyz9Uq0=
Message-ID: <dcc563d10809011342i55d32d23la4604e71d19dfdb0@mail.gmail.com>
Date: Mon, 1 Sep 2008 14:42:07 -0600
From: "Scott Marlowe" <scott.marlowe@gmail.com>
To: "Thomas Finneid" <tfinneid@student.matnat.uio.no>
Subject: Re: slow update of index during insert/copy
Cc: "Scott Carey" <scott@richrelevance.com>,
	"Craig Ringer" <craig@postnewspapers.com.au>,
	pgsql-performance@postgresql.org
In-Reply-To: <48BC5178.60906@ifi.uio.no>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
References: <48BA9D65.2020001@ifi.uio.no>
	<48BB4A2D.9080404@postnewspapers.com.au> <48BBD22E.4090308@ifi.uio.no>
	<dcc563d10809011051s558f8a7bp5beb0dbe181721e4@mail.gmail.com>
	<a1ec7d000809011146x456b942eof33627c63f8a021a@mail.gmail.com>
	<48BC5178.60906@ifi.uio.no>

On Mon, Sep 1, 2008 at 2:32 PM, Thomas Finneid
<tfinneid@student.matnat.uio.no> wrote:
>
> Scott Carey wrote:
>>
>> For a development box, just enable write-back caching regardless of the
>> battery back up situation.  As long as its not your only copy of
>
> Will have a look at it, the data is not important and can be reproduced any
> time on any machine. The controller I have is a Areca ARC-1220 Serial ATA 8
> port RAID Controller - PCI-E, SATA II, so I dont know exactly what it
> supports of caching.

It's a pretty good card.  It should support 1G of cache at least, and
definitely supports battery backup.  Have had a pair of 1680 Arecas in
production for a month now and so far I'm very happy with the
reliability and performance.

The other Scott is technically right about the lack of need for
battery back on a dev machine as long as you go in and change the
chaching to write back, and I'm sure the card will give you a red
dialog box saying this is a bad idea. Now, if it would take you a day
of downtime to get a dev database back in place and running after a
power loss, then the bbu may be worth the $200 or so.

While I like to have a machine kitted out just like prod for testing,
for development I do tend to prefer slower machines, so that
developers might notice if they've just laid a big fat bloaty code
egg that's slower than it should be.  Try to optimize for acceptable
performance on such server and you're far more likely to get good
performance behaviour in production.