X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (av.hub.org [200.46.204.144]) by postgresql.org (Postfix) with ESMTP id 4461A9DCD76 for ; Thu, 1 Dec 2005 15:19:37 -0400 (AST) Received: from postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36925-06 for ; Thu, 1 Dec 2005 15:19:33 -0400 (AST) X-Greylist: from auto-whitelisted by SQLgrey- Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by postgresql.org (Postfix) with ESMTP id 283809DCB7B for ; Thu, 1 Dec 2005 15:19:33 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id jB1JJEHq015961; Thu, 1 Dec 2005 14:19:14 -0500 (EST) To: "Jeremy Haile" cc: pgsql-performance@postgresql.org Subject: Re: Insert performance slows down in large batch In-reply-to: <1133459351.23834.248779098@webmail.messagingengine.com> References: <1133459351.23834.248779098@webmail.messagingengine.com> Comments: In-reply-to "Jeremy Haile" message dated "Thu, 01 Dec 2005 12:49:11 -0500" Date: Thu, 01 Dec 2005 14:19:14 -0500 Message-ID: <15960.1133464754@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, score=0 required=5 tests=[none] X-Spam-Score: 0 X-Spam-Level: X-Archive-Number: 200512/39 X-Sequence-Number: 15860 "Jeremy Haile" writes: > 1) Why does the performance degrade as the table sizes grow? Shouldn't > the insert performance remain fairly constant if there are no indexes or > constraints? Yeah, insert really should be a constant-time operation if there's no add-on operations like index updates or FK checks. Can you get more information about where the time is going with gprof or oprofile? (I'm not sure if oprofile is available for RHEL4, but it is in Fedora 4 so maybe RHEL4 has it too.) If you're not comfortable with performance measurement tools, perhaps you could crank up a test case program that just generates dummy data and inserts it in the same way as your real application does. If you can confirm a slowdown in a test case that other people can look at, we'd be happy to look into the reason for it. regards, tom lane