X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (av.hub.org [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 83331DA0F0 for ; Tue, 4 Oct 2005 22:09:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 49087-05 for ; Wed, 5 Oct 2005 01:09:02 +0000 (GMT) Received: from mail.baymountain.com (mail.baymountain.com [8.7.96.19]) by svr1.postgresql.org (Postfix) with SMTP id 2E462DA0AB for ; Tue, 4 Oct 2005 22:09:03 -0300 (ADT) Received: (qmail 30517 invoked by uid 504); 5 Oct 2005 01:09:06 -0000 Received: from emil@baymountain.com by mail.baymountain.com by uid 501 with qmail-scanner-1.16 (spamassassin: 2.53. Clear:SA:0(0.0/7.0):. Processed in 0.266532 secs); 05 Oct 2005 01:09:06 -0000 Received: from unknown (HELO briggspack.com) (emil@briggspack.com@24.211.148.77) by mail.baymountain.com with SMTP; 5 Oct 2005 01:09:06 -0000 From: Emil Briggs Reply-To: emil@baymountain.com Organization: Baymountain, Inc. To: pgsql-performance@postgresql.org Subject: Indexes on ramdisk Date: Tue, 4 Oct 2005 21:09:01 -0400 User-Agent: KMail/1.7.1 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200510042109.01849.emil@baymountain.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.037 required=5 tests=[AWL=-0.013, FORGED_RCVD_HELO=0.05] X-Spam-Level: X-Archive-Number: 200510/63 X-Sequence-Number: 14831 I have an application that has a table that is both read and write intensive. Data from iostat indicates that the write speed of the system is the factor that is limiting performance. The table has around 20 columns and most of the columns are indexed. The data and the indices for the table are distributed over several mirrored disk partitions and pg_xlog is on another. I'm looking at ways to improve performance and besides the obvious one of getting an SSD I thought about putting the indices on a ramdisk. That means that after a power failure or shutdown I would have to recreate them but that is acceptable for this application. What I am wondering though is whether or not I would see much performance benefit and if there would be any startup problems after a power down event due to the indices not being present. Any insight would be appreciated. Emil