From owner-freebsd-isp Thu Jul 6 11: 2:13 2000 Delivered-To: freebsd-isp@freebsd.org Received: from Samizdat.uucom.com (samizdat.uucom.com [198.202.217.54]) by hub.freebsd.org (Postfix) with ESMTP id 63C6137BAFB for ; Thu, 6 Jul 2000 11:02:06 -0700 (PDT) (envelope-from cshenton@uucom.com) Received: (from cshenton@localhost) by Samizdat.uucom.com (8.9.3/8.9.3) id OAA00515; Thu, 6 Jul 2000 14:01:51 -0400 (EDT) To: Luigi Rizzo Cc: Alan Batie , isp@FreeBSD.ORG Subject: Re: load balancing References: <200007061641.SAA85813@info.iet.unipi.it> From: Chris Shenton Date: 06 Jul 2000 14:01:51 -0400 In-Reply-To: Luigi Rizzo's message of "Thu, 6 Jul 2000 18:41:46 +0200 (CEST)" Message-ID: Lines: 15 User-Agent: Gnus/5.0807 (Gnus v5.8.7) Emacs/20.4 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-freebsd-isp@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, 6 Jul 2000 18:41:46 +0200 (CEST), Luigi Rizzo said: Luigi> I have some plans to implement a load balancer a-la Cisco Luigi> LocalDirector for FreeBSD, but this requires one machine to sit Luigi> between clients and servers. Problem with having a single machine as the balancer is that you inject a single point of failure. IMHO the balancer must support some kind of failover to a twin balancer box, or better, share load between the two balancers if they're both up. DJB's djbdns (aka "dnscache" suite) claims to do something like this with DNS but I don't believe it checks server health at the application layer, which IMHO it must if it's to be used as for fault tolerance at minimum, balancing at best. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message