From owner-freebsd-questions Thu Sep 11 17:30:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id RAA12031 for questions-outgoing; Thu, 11 Sep 1997 17:30:49 -0700 (PDT) Received: from tok.qiv.com ([204.214.141.211]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id RAA12026 for ; Thu, 11 Sep 1997 17:30:46 -0700 (PDT) Received: (from uucp@localhost) by tok.qiv.com (8.8.6/8.8.5) with UUCP id TAA18260; Thu, 11 Sep 1997 19:30:32 -0500 (CDT) Received: from localhost (jdn@localhost) by acp.qiv.com (8.8.6/8.8.5) with SMTP id TAA00556; Thu, 11 Sep 1997 19:27:40 -0500 (CDT) X-Authentication-Warning: acp.qiv.com: jdn owned process doing -bs Date: Thu, 11 Sep 1997 19:27:40 -0500 (CDT) From: "Jay D. Nelson" To: Steven Ames cc: freebsd-questions@FreeBSD.ORG Subject: Re: Clustering/fail-over capability? In-Reply-To: <199709111913.OAA07761@news.cioe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-questions@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Thu, 11 Sep 1997, Steven Ames wrote: > > > Mirroring disks is pretty easy to do, the built-in disk striping driver > > (ccd) does this already. But the rest is pretty tricky. > > But in this case you'd really want to be mirroring the disks across servers, > not within the same server. ccd can't do that. No -- you don't want to mirror between servers. Put _two_ differential controllers in both machines and _two_ external SCSI towers cabeled between the two machines. Mirror disks on one controller to the disks on the other controller. The takeover is reasonably complicated -- managing machine and process state as well as outstanding locks and terminal sessions is non-trivial in an exclusive takeover situation. Concurrent access to shared disks, which you would need for load balancing is even worse. I think this is way beyond the scope of reasonable expectations for a volunteer effort. While many have the skills -- and maybe the desire -- groceries are more important. Commercial software that does this goes for $10-30K. Maybe you could hone your skills (and future income) and implement this for us? -- Jay > > I believe that 'clustering' is going to become a more important topic in > the near future. Having fail over and load balanced servers is desired for > larger 'mission critical' applications. > > > -Steve