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 C9D10D1C4B8 for ; Tue, 29 Jul 2003 15:46:18 +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 76219-07 for ; Tue, 29 Jul 2003 12:45:18 -0300 (ADT) Received: from lakemtao02.cox.net (lakemtao02.cox.net [68.1.17.243]) by svr1.postgresql.org (Postfix) with ESMTP id CD902D1C4B7 for ; Tue, 29 Jul 2003 12:46:16 -0300 (ADT) Received: from [192.168.1.1] ([68.11.66.83]) by lakemtao02.cox.net (InterMail vM.5.01.04.05 201-253-122-122-105-20011231) with ESMTP id <20030729154616.SXAG24359.lakemtao02.cox.net@[192.168.1.1]> for ; Tue, 29 Jul 2003 11:46:16 -0400 Subject: Re: Tuning PostgreSQL From: Ron Johnson To: PgSQL Performance ML In-Reply-To: References: <873cgvpylu.fsf@stark.dyndns.tv> Content-Type: text/plain Message-Id: <1059493575.7505.129.camel@haggis> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.3 Date: 29 Jul 2003 10:46:15 -0500 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200307/465 X-Sequence-Number: 2758 On Tue, 2003-07-29 at 10:14, Vivek Khera wrote: > >>>>> "GS" == Greg Stark writes: > > GS> "scott.marlowe" writes: > > GS> But you have to actually test your setup in practice to see if it > GS> hurts. A big data warehousing system will be faster under RAID5 > GS> than under RAID1+0 because of the extra disks in the > GS> stripeset. The more disks in the stripeset the more bandwidth you > GS> get. > > Anyone have ideas on 14 spindles? I just ordered a disk subsystem > with 14 high speed (U320 15kRPM) SCSI disks to hook up with a dell > PERC3/DC controller (only 128MB cache, though). 14 drives on one SCSI card, eh? I'd be worried about saturating the bus. Maybe it's an old rule of thumb, but I would fill a SCSI chain more than half full. > My plan was to do RAID10, but I think I'll play with RAID5 again and > see which gives me the best performance. Unfortunatly, it is > difficult to recreate the highly fragmented tables I have now (vacuum > full takes over 14 hours on one of the tables) so I'm not sure how to > best compare them. Also IMO: if I needed something *really* high performance, I'd start with a mobo that has dual PCI buses (133MB/s can get swamped quickly by U320 devices) or PCI-X (but they're so new...). Then: - get dual U320 SCSI cards (one for each PCI bus) - plug them into dual redundant fast path external storage controllers that have, oh, 512MB RAM cache *each*) - dual port the drives, so they plug into both storage controllers Since I wouldn't put 14 drives on one SCSI chain, double what I just said, and only plug 7 drives in each controller. If your app needs One Big Honkin' Device, use the Linux Volume Manager (LVM) to merge the 2 RAID logical devices into one "super- logical" device. Yes, that's lot's of money, but is the data, and speed, important enough? -- +-----------------------------------------------------------------+ | 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 | +-----------------------------------------------------------------+