X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2D9FAD1B445 for ; Wed, 3 Dec 2003 19:33:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88160-01 for ; Wed, 3 Dec 2003 15:32:51 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 3681FD1B4D6 for ; Wed, 3 Dec 2003 15:32:50 -0400 (AST) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id E4EAF37372; Wed, 3 Dec 2003 14:32:50 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1ARcjy-00022F-00; Wed, 03 Dec 2003 14:32:50 -0500 To: Jared Carr Cc: Tom Lane , Greg Stark , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> <26699.1070423607@sss.pgh.pa.us> <2704.1070473746@sss.pgh.pa.us> <3FCE2B08.6090606@89glass.com> In-Reply-To: <3FCE2B08.6090606@89glass.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 03 Dec 2003 14:32:50 -0500 Message-ID: <87k75d66jh.fsf@stark.dyndns.tv> Lines: 17 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/52 X-Sequence-Number: 4912 Jared Carr writes: > The patch definitely makes things more consistent...unfortunately it is more > consistent toward the slower execution times. Of course I am looking at this > simply from a straight performance standpoint and not a viewpoint of what > *should* be happening. At any rate here are the query plans with the various > settings. The optimizer seems to be at least considering reasonable plans now. It seems from the estimates that you need to rerun analyze. You might try "vacuum full analyze" to be sure. Also, you might try raising effective_cache_size and/or lowering random_page_size (it looks like something around 2 might help). -- greg