Received: from localhost (unknown [200.46.204.187]) by postgresql.org (Postfix) with ESMTP id DF0A89F972A for ; Mon, 22 Oct 2007 07:04:38 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.187]) (amavisd-maia, port 10024) with ESMTP id 25428-03 for ; Mon, 22 Oct 2007 07:04:29 -0300 (ADT) X-Greylist: delayed 00:19:47.733315 by SQLgrey-1.7.5 Received: from pjetlic.mobart.hr (pjetlic.mobart.hr [80.80.51.10]) by postgresql.org (Postfix) with ESMTP id 9F2C59F9A90 for ; Mon, 22 Oct 2007 07:04:31 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by pjetlic.mobart.hr (Postfix) with ESMTP id 295693CAE1; Mon, 22 Oct 2007 11:44:35 +0200 (CEST) Received: from pjetlic.mobart.hr ([127.0.0.1]) by localhost (pjetlic [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 12929-12; Mon, 22 Oct 2007 11:44:28 +0200 (CEST) Received: from [192.168.1.94] (78-1-99-203.adsl.net.t-com.hr [78.1.99.203]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by pjetlic.mobart.hr (Postfix) with ESMTP id 507073CADB; Mon, 22 Oct 2007 11:44:28 +0200 (CEST) Message-ID: <471C70F9.7060407@megafon.hr> Date: Mon, 22 Oct 2007 11:44:25 +0200 From: Mario Splivalo User-Agent: Thunderbird 1.5.0.13 (X11/20070824) MIME-Version: 1.0 To: cho Cc: pgsql-admin@postgresql.org Subject: Re: Cluster/Replication References: In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by OpenSource AV+AS Mail Server X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200710/208 X-Sequence-Number: 27122 cho wrote: > Le Fri, 19 Oct 2007 12:43:40 -0400, Sean Brown a écrit : > > Have a look at pgpool-I or pgpool-II in replication mode. > Pgpool proxies SQL requests, so that requests on one server are replayed > on the other. pgpool-I is limited to two nodes, whereas pgpool-II is not. > Latency is important for write performance. You can do failover, you just > need to detect by yourself when one node is dead. > > There is also PITR recovery, but as it batches the transaction logs you > may lose a bit of data (up to a configured data size / duration). > > PITR does not allow for the 'slave' server to be on-line, meaning you can't do SELECTs on the slave while shipping WALs. Mike