From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Clements Subject: Re: active/active vs active/passive? Date: Wed, 08 Jun 2005 13:51:03 -0400 Message-ID: <42A73007.6060102@steeleye.com> References: <1118189397.15459.108.camel@seki.nac.uci.edu> <42A641CC.9040905@steeleye.com> <1118250910.15459.132.camel@seki.nac.uci.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1118250910.15459.132.camel@seki.nac.uci.edu> Sender: linux-raid-owner@vger.kernel.org To: Dan Stromberg Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids Dan Stromberg wrote: > On Tue, 2005-06-07 at 20:54 -0400, Paul Clements wrote: > What it is about active/passive that is so much easier to implement than > active/active? There's nothing that difficult about active/active, assuming you're talking about the servers operating on separate (or static) data. What's difficult is the cluster filesystem. Those tend to be big pieces of low-level (read, kernel) code, and they require a distributed lock manager and a membership model, which are also generally implemented in-kernel. Of course, without knowing the context of the speaker's reference to this, it's hard to say exactly what he meant or if it's even remotely related to what I'm talking about... :) -- Paul