Received: from localhost (unknown [200.46.204.183])
	by mail.postgresql.org (Postfix) with ESMTP id CD7D96501D1
	for <pgsql-performance-postgresql.org@mail.postgresql.org>;
	Wed, 10 Dec 2008 20:08:34 -0400 (AST)
Received: from mail.postgresql.org ([200.46.204.86])
	by localhost (mx1.hub.org [200.46.204.183]) (amavisd-maia, port 10024)
	with ESMTP id 53140-04
	for <pgsql-performance-postgresql.org@mail.postgresql.org>;
	Wed, 10 Dec 2008 20:08:31 -0400 (AST)
X-Greylist: from auto-whitelisted by SQLgrey-1.7.6
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by mail.postgresql.org (Postfix) with ESMTP id 865A865032E
	for <pgsql-performance@postgresql.org>;
	Wed, 10 Dec 2008 20:08:31 -0400 (AST)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.14.2/8.14.2) with ESMTP id mBB08S2X009071;
	Wed, 10 Dec 2008 19:08:28 -0500 (EST)
To: Richard Yen <dba@richyen.com>
cc: Robert Haas <robertmhaas@gmail.com>, pgsql-performance@postgresql.org
Subject: Re: query plan with index having a btrim is different for strings of
	different length
In-reply-to: <129E8D79-341B-411A-83B7-E2B509A7B6E2@richyen.com> 
References: <751F49DC-1142-4E9C-95C8-BC3C229B3F0B@richyen.com>
	<22937.1228865243@sss.pgh.pa.us>
	<E8AE9590-4E30-49C0-B11F-D7131009A095@richyen.com>
	<603c8f070812101139p189cc828hef8702f37f952b6d@mail.gmail.com>
	<129E8D79-341B-411A-83B7-E2B509A7B6E2@richyen.com>
Comments: In-reply-to Richard Yen <dba@richyen.com>
	message dated "Wed, 10 Dec 2008 14:04:57 -0800"
Date: Wed, 10 Dec 2008 19:08:28 -0500
Message-ID: <9070.1228954108@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>
X-Virus-Scanned: Maia Mailguard 1.0.1
X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests=none
X-Spam-Level: 
X-Archive-Number: 200812/132
X-Sequence-Number: 32168

Richard Yen <dba@richyen.com> writes:
> Is there any way to tune this so that for the common last names, the query  
> run time doesn't jump from <1s to >300s?

Well, as near as I can tell there's factor of a couple hundred
difference between the frequencies of 'smith' and 'smithers', so
you shouldn't really expect similar runtimes for the two cases.

Having said that, I still think you should try to index both first
and last name.  Also I wonder whether the index on owner is worth
having at all.  It definitely doesn't seem worthwhile to index the
entries with owner = -1, since there are so many; so maybe you could
make it a partial index that excludes those entries, in order to prevent
the planner from trying to use it for this case.

			regards, tom lane