Received: from mblox.com (unknown [213.155.153.59]) by postgresql.org (Postfix) with ESMTP id CD0D54759AA for ; Tue, 23 Apr 2002 10:09:19 -0400 (EDT) Received: from salamander (salamander.mblox.com [::ffff:192.168.2.102]) (AUTH: LOGIN ian.cass) by aerosmith with esmtp; Tue, 23 Apr 2002 14:09:23 +0000 Message-ID: <003001c1ead0$41169260$6602a8c0@salamander> From: "Ian Cass" To: "Dima Tkach" Cc: pgsql-sql@postgresql.org References: <005501c1eaaa$46eb2500$6602a8c0@salamander> <3CC559FF.1030209@openratings.com> Subject: Re: Date indexing Date: Tue, 23 Apr 2002 15:07:49 +0100 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2600.0000 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 X-Archive-Number: 200204/277 X-Sequence-Number: 7541 > This won't be able to make much use of the compound index, because you > are not searching for exact match on the first column, and the other > columns are not in your criteria at all ... Thus your index is only as > good as being just on the timestamp alone. This makes sense. I'll change the order of my indexing. > 10 minutes still sounds pretty excessive though, even for a seq. scan > on a 5000 rows table. Have you vacuum'ed it lately? > Try running 'vacuum full' on that table, it might help a lot, if you > were ding lots of updates to it. No, the actual table contains more like 10mill lines, however, the dataset I'd be referencing if the index were working as I intended would be a mere 5k rows. Thanks for your help. I'll try this later today. -- Ian Cass