X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [64.117.224.130]) by svr1.postgresql.org (Postfix) with ESMTP id 87F88D1C4B8 for ; Tue, 29 Jul 2003 20:38:36 +0000 (GMT) Received: from svr1.postgresql.org ([64.117.224.193]) by localhost (neptune.hub.org [64.117.224.130]) (amavisd-new, port 10024) with ESMTP id 97315-02 for ; Tue, 29 Jul 2003 17:37:38 -0300 (ADT) Received: from lakemtao03.cox.net (lakemtao03.cox.net [68.1.17.242]) by svr1.postgresql.org (Postfix) with ESMTP id 82BDFD1C4D5 for ; Tue, 29 Jul 2003 17:38:34 -0300 (ADT) Received: from [192.168.1.1] ([68.11.66.83]) by lakemtao03.cox.net (InterMail vM.5.01.04.05 201-253-122-122-105-20011231) with ESMTP id <20030729203835.XRQW23518.lakemtao03.cox.net@[192.168.1.1]>; Tue, 29 Jul 2003 16:38:35 -0400 Subject: Re: Tuning PostgreSQL From: Ron Johnson To: "scott.marlowe" Cc: PgSQL Performance ML In-Reply-To: References: Content-Type: text/plain Message-Id: <1059511115.7508.197.camel@haggis> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.3 Date: 29 Jul 2003 15:38:35 -0500 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200307/473 X-Sequence-Number: 2766 On Tue, 2003-07-29 at 15:09, scott.marlowe wrote: > On 29 Jul 2003, Ron Johnson wrote: > > > On Tue, 2003-07-29 at 14:00, scott.marlowe wrote: > > > On 29 Jul 2003, Ron Johnson wrote: > > > > > > > On Tue, 2003-07-29 at 11:18, scott.marlowe wrote: > > > > > On 29 Jul 2003, Ron Johnson wrote: > > > > > > > > > > > On Tue, 2003-07-29 at 10:14, Vivek Khera wrote: > > > > > > > >>>>> "GS" == Greg Stark writes: > > > > > > > > > > > > > > GS> "scott.marlowe" writes: [snip] > > > But that's seq scan. For many database applications, random access > > > performance is much more important. Imagine 200 people entering > > > reservations of 8k or less each into a transaction processing engine. > > > Each transactions chance to hit an unoccupied spindle is what really > > > counts. If there's 30 spindles, each doing a stripe's worth of access all > > > the time, it's likely to never flood the channel. > > > > > > If random access is 1/4th the speed of seq scan, then you need to multiply > > > it by 4 to get the number of drives that'd saturate the PCI bus. > > > > Maybe it's just me, but I've never seen a purely TP system. > > I think most of them are running under TPF on a mainframe in a basement > somewhere, like for airline reservations. I've never worked on one, but > met one of the guys who runs one, and they use 12 mainframes for 6 live > machines and each live machine has a failover machine behind it in sysplex > mode. I kept thinking of the giant dinosaurs in Jurassic park... We have something similar running on Alphas and VMS; does about 8M Txn/day. Anyone who uses E-ZPass in the northeast eventually gets stuck in our systems. (Made me fear Big Brother...) > > Even if roll off the daily updates to a "reporting database" each > > night, some yahoo manager with enough juice to have his way still > > wants up-to-the-minute reports... > > Just because it's TP doesn't mean it doesn't have real time reporting. > But expensive reports probably do get run at night. Yes, but... There's always the exception. > > Better yet, the Access Jockey, who thinks s/he's an SQL whiz but > > couldn't JOIN himself out of a paper bag... > > I've seen a few who got joins and unions and what not, but explaining fks > or transactions got me a glazed look... :-) Wow! They understood joins? You lucky dog!!! -- +-----------------------------------------------------------------+ | Ron Johnson, Jr. Home: ron.l.johnson@cox.net | | Jefferson, LA USA | | | | "I'm not a vegetarian because I love animals, I'm a vegetarian | | because I hate vegetables!" | | unknown | +-----------------------------------------------------------------+