X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 06AA28BA349 for ; Tue, 15 Feb 2005 20:16:29 +0000 (GMT) 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 11011-07 for ; Tue, 15 Feb 2005 20:16:24 +0000 (GMT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 17CD58BA179 for ; Tue, 15 Feb 2005 20:16:24 +0000 (GMT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.13.1/8.13.1) with ESMTP id j1FKGJR9039680 for ; Tue, 15 Feb 2005 20:16:21 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.13.1/8.13.1/Submit) id j1FJoBmn034132 for pgsql-performance@postgresql.org; Tue, 15 Feb 2005 19:50:11 GMT (envelope-from news) From: Christopher Browne X-Newsgroups: pgsql.performance Subject: Re: seq scan cache vs. index cache smackdown Date: Tue, 15 Feb 2005 14:04:49 -0500 Organization: cbbrowne Computing Inc Lines: 22 Message-ID: References: <6EE64EF3AB31D5448D0007DD34EEB3412A761E@Herge.rcsinc.local> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@news.hub.org X-message-flag: Outlook is rather hackable, isn't it? X-Home-Page: http://www.cbbrowne.com/info/ X-Affero: http://svcs.affero.net/rm.php?r=cbbrowne User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4 (Corporate Culture, linux) Cancel-Lock: sha1:P8ZOfb1UHdusu/2WiuQ1+RJA/yU= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200502/244 X-Sequence-Number: 10573 In the last exciting episode, merlin.moncure@rcsonline.com ("Merlin Moncure") wrote: >> It seems inevitable that Postgres will eventually eliminate that >> redundant layer of buffering. Since mmap is not workable, that >> means using O_DIRECT to read table and index data. > > What about going the other way and simply letting the o/s do all the > caching? How bad (or good) would the performance really be? I'm going to see about taking this story to OLS (Ottawa Linux Symposium) in July and will see what hearing I can get. There are historically some commonalities in the way this situation is regarded, in that there was _long_ opposition to the notion of having unbuffered disk devices. If there's more "story" that definitely needs to be taken, let me know... -- output = reverse("moc.enworbbc" "@" "enworbbc") http://linuxdatabases.info/info/slony.html Rules of the Evil Overlord #90. "I will not design my Main Control Room so that every workstation is facing away from the door."