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 0A9F63A55A3 for ; Tue, 25 Jan 2005 13:35:54 +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 58584-09 for ; Tue, 25 Jan 2005 13:35:42 +0000 (GMT) Received: from fuji.krosing.net (joltid-gw.joltid.org [195.50.194.24]) by svr1.postgresql.org (Postfix) with ESMTP id 900363A5548 for ; Tue, 25 Jan 2005 13:35:44 +0000 (GMT) Received: from fuji.krosing.net (localhost [127.0.0.1]) by fuji.krosing.net (8.13.1/8.13.1) with ESMTP id j0PAgpL6010732; Tue, 25 Jan 2005 12:42:51 +0200 Received: (from hannu@localhost) by fuji.krosing.net (8.13.1/8.13.1/Submit) id j0PAgmMv010731; Tue, 25 Jan 2005 12:42:48 +0200 X-Authentication-Warning: fuji.krosing.net: hannu set sender to hannu@tm.ee using -f Subject: Re: PostgreSQL clustering VS MySQL clustering From: Hannu Krosing To: Tatsuo Ishii Cc: tgl@sss.pgh.pa.us, marty@outputservices.com, herve@elma.fr, pgsql-performance@postgresql.org In-Reply-To: <20050124.115244.35008937.t-ishii@sra.co.jp> References: <41F14758.4030602@outputservices.com> <20050122.121300.41631131.t-ishii@sra.co.jp> <25335.1106428223@sss.pgh.pa.us> <20050124.115244.35008937.t-ishii@sra.co.jp> Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: quoted-printable Date: Tue, 25 Jan 2005 12:42:47 +0200 Message-Id: <1106649767.5790.18.camel@fuji.krosing.net> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200501/458 X-Sequence-Number: 10148 =DChel kenal p=E4eval (esmasp=E4ev, 24. jaanuar 2005, 11:52+0900), kirjutas Tatsuo Ishii: > > Tatsuo Ishii writes: > > > Probably VACUUM works well for small to medium size tables, but not > > > for huge ones. I'm considering about to implement "on the spot > > > salvaging dead tuples". > >=20 > > That's impossible on its face, except for the special case where the > > same transaction inserts and deletes a tuple. In all other cases, the > > transaction deleting a tuple cannot know whether it will commit. >=20 > Of course. We need to keep a list of such that tuples until commit or > abort. what about other transactions, which may have started before current one and be still running when current one commites ? I once proposed an extra parameter added to VACUUM FULL which determines how much free space to leave in each page vacuumed. If there were room the new tuple could be placed near the old one in most cases and thus avoid lots of disk head movement when updating huge tables in one go. ------------ Hannu Krosing