X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 18542D1E137 for ; Mon, 12 Apr 2004 21:17:35 -0300 (ADT) 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 47709-01 for ; Mon, 12 Apr 2004 21:17:36 -0300 (ADT) Received: from mail4.speakeasy.net (mail4.speakeasy.net [216.254.0.204]) by svr1.postgresql.org (Postfix) with ESMTP id 9866ED1E129 for ; Mon, 12 Apr 2004 21:17:32 -0300 (ADT) Received: (qmail 10051 invoked by uid 64014); 13 Apr 2004 00:17:36 -0000 Received: from kgeis@speakeasy.org by mail4.speakeasy.net with AmikaGuardian-Server-2.1.2; 13 Apr 2004 00:17:36 -0000 X-AmikaGuardian-Id: mail4.speakeasy.net108181545526110043 X-AmikaGuardian-Action: Do Nothing() Received: from adsl-67-113-18-226.dsl.snfc21.pacbell.net (HELO speakeasy.org) (kgeis@[67.113.18.226]) (envelope-sender ) by mail4.speakeasy.net (qmail-ldap-1.03) with AES256-SHA encrypted SMTP for ; 13 Apr 2004 00:17:35 -0000 Message-ID: <407B319B.7000400@speakeasy.org> Date: Mon, 12 Apr 2004 17:17:31 -0700 From: Ken Geis User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7b) Gecko/20040316 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Tracking down performance issue Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/144 X-Sequence-Number: 6444 I'm running 7.4.2 on an older Linux box (450MHzAMD K-6-III, 450M RAM) running kernel 2.6.5. My client is a Java/JDBC program on Windows. I'm having trouble seeing where the bottleneck in my performance is. The client uses about 30% CPU. The server uses 70% CPU plus 1.5% I/O wait. The I/O wait is very low because I'm doing a PK index scan where the index and data are on different disks and the table is clustered on the PK index. The network is 100Mb, and it's at 7% of capacity. I tried making the client operate on two threads on two database connections. That bumped the server utilization to 80% and barely changed the I/O wait. The throughput decreased by a third. The only thing I can think of is memory bandwidth. Does anyone have tips on how I can investigate more?