X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B88E5D1B4FD for ; Thu, 23 Oct 2003 06:15:34 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89710-01 for ; Thu, 23 Oct 2003 03:15:04 -0300 (ADT) Received: from relay.icomedias.com (relay.icomedias.com [62.99.232.66]) by svr1.postgresql.org (Postfix) with ESMTP id C245CD1B4E6 for ; Thu, 23 Oct 2003 03:15:01 -0300 (ADT) Received: from loki.icomedias.com ([10.192.17.128]) by relay.icomedias.com (8.12.10/8.12.10) with ESMTP id h9N6Eu5A026552 for ; Thu, 23 Oct 2003 08:14:56 +0200 From: Mario Weilguni To: pgsql-performance@postgresql.org Subject: Re: vacuum locking Date: Thu, 23 Oct 2003 08:14:56 +0200 User-Agent: KMail/1.5.4 References: <16272.318.870000.93584@gargle.gargle.HOWL> <16279.4988.496000.539525@gargle.gargle.HOWL> In-Reply-To: <16279.4988.496000.539525@gargle.gargle.HOWL> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200310230814.56738.mweilguni@sime.com> X-Scanned-By: MIMEDefang 2.33 (www . roaringpenguin . com / mimedefang) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200310/594 X-Sequence-Number: 4342 Am Donnerstag, 23. Oktober 2003 01:32 schrieb Rob Nagler: > The concept of vacuuming seems to be problematic. I'm not sure why > the database simply can't garbage collect incrementally. AGC is very > tricky, especially AGC that involves gigabytes of data on disk. > Incremental garbage collection seems to be what other databases do, > and it's been my experience that other databases don't have the type > of unpredictable behavior I'm seeing with Postgres. I'd rather the > database be a little bit slower on average than have to figure out the > best time to inconvenience my users. I think oracle does not do garbage collect, it overwrites the tuples directly and stores the old tuples in undo buffers. Since most transactions are commits, this is a big win.