Date: Sun, 19 Aug 2007 11:41:15 -0700
From: David Fetter <david@fetter.org>
To: mark overmeer <markovermeer@gmail.com>,
	pgsql-performance@postgresql.org
Subject: Re: schema design question
Message-ID: <20070819184115.GL28490@fetter.org>
References: <ea41f7570708190619u6daf7cep96430d20566d6159@mail.gmail.com>
	<20070819181216.GK28490@fetter.org> <20070819182658.GC25214@uio.no>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20070819182658.GC25214@uio.no>
User-Agent: Mutt/1.5.14 (2007-02-12)

On Sun, Aug 19, 2007 at 08:26:58PM +0200, Steinar H. Gunderson wrote:
> On Sun, Aug 19, 2007 at 11:12:16AM -0700, David Fetter wrote:
> > There's your mistake.  EAV is not performant, and won't become so.
> 
> It sort of depends. I put all the EXIF information for my image
> gallery into an EAV table -- it was the most logical format at the
> time, although I'm not sure I need all the information. Anyhow, with
> clustering and indexes, Postgres zips through the five million
> records easily enough for my use -- at least fast enough that I can
> live with it without feeling the need for a redesign.

Unless your records are huge, that's a tiny database, where tiny is
defined to mean that the whole thing fits in main memory with plenty
of room to spare.  I guarantee that performance will crash right
through the floor as soon as any table no longer fits in main memory.

> As a general database design paradigm, though, I fully agree with
> you.  Databases are databases, not glorified OO data stores or hash
> tables.

Exactly :)

Cheers,
David.
-- 
David Fetter <david@fetter.org> http://fetter.org/
phone: +1 415 235 3778        AIM: dfetter666
                              Skype: davidfetter

Remember to vote!
Consider donating to PostgreSQL: http://www.postgresql.org/about/donate