Received: from localhost (unknown [200.46.204.184]) by postgresql.org (Postfix) with ESMTP id 7E57C9F9F47 for ; Mon, 22 Oct 2007 05:40:15 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.184]) (amavisd-maia, port 10024) with ESMTP id 59410-05 for ; Mon, 22 Oct 2007 05:40:07 -0300 (ADT) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.5 Received: from ciao.gmane.org (main.gmane.org [80.91.229.2]) by postgresql.org (Postfix) with ESMTP id 33DD69F9CE7 for ; Mon, 22 Oct 2007 05:40:11 -0300 (ADT) Received: from root by ciao.gmane.org with local (Exim 4.43) id 1IjspO-0001cc-R9 for pgsql-admin@postgresql.org; Mon, 22 Oct 2007 08:40:02 +0000 Received: from dhcp26-237.enst.fr ([137.194.26.237]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Mon, 22 Oct 2007 08:40:02 +0000 Received: from tobutaz by dhcp26-237.enst.fr with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Mon, 22 Oct 2007 08:40:02 +0000 X-Injected-Via-Gmane: http://gmane.org/ To: pgsql-admin@postgresql.org From: cho Subject: Re: Cluster/Replication Date: Mon, 22 Oct 2007 08:34:24 +0000 (UTC) Lines: 22 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-Complaints-To: usenet@ger.gmane.org X-Gmane-NNTP-Posting-Host: dhcp26-237.enst.fr User-Agent: Pan/0.129 (Benson & Hedges Moscow Gold) X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200710/205 X-Sequence-Number: 27119 Le Fri, 19 Oct 2007 12:43:40 -0400, Sean Brown a écrit : > We are setting up a new database system that is going to act as our > central database. We are planning to have a Master/Slave setup, > originally with one slave but there is a strong possibility of adding a > second slave at our second site. We could run Mulit-Master, it just > wasn't the original plan. (The original plan was MySQL log shipping). We > do not want the possibility of loosing data in the event of loosing the > master node so synchronous replication is preferable. We are using > BLOBS, so the replication has to be able to handle that. The slave we > want to have useable but it would be used for reporting only under > normal usage. 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).