Date: Wed, 21 Nov 2001 10:41:23 +1100 (EST) From: Enno Davids <enno.davids@metva.com.au> To: shaun@shamz.net (Shaun Jurrens) Cc: freebsd-isp@FreeBSD.ORG Subject: Re: Clustering software Message-ID: <200111202341.KAA02627@metva.com.au> In-Reply-To: <20011120094227.G24849@atreides.freenix.no> from Shaun Jurrens at "Nov 20, 1 09:42:27 am"
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 20 Nov 2001 09:42:27 +0100 Shaun Jurrens <shaun@shamz.net> wrote: | #>Hi, | #> | #>Unfortunately the searching I've done yields little or nothing, especially | #>if you are talking high-availability-type clustering. Seems the linux | #>people are the only ones doing any serious work on that - see | #>www.linux-ha.org. You might find some links to, or actual software, you | #>can use/port. | | Unfortunately this is not at all the case. I don't know where you've | done your searching, but it hasn't been very thorough. Ronald Minnich | (excuse any bad spelling) has been doing clustering with FreeBSD for _years_ | and has clusters of 64 and more (iirc) machines. I'd appreciate if you did | better research before you post such information to the lists. It only | serves to disinform those who search in the list archives. I suspect the discussion here has diverged. Clearly Shaun and the first respondant (who's name I can't see here - my apologies) are talking about different sorts of clustering. The large clusters of 64 fairly clearly are Beowulf style systems where a single task is broken down and sprayd across many systems to gain processing advantage out of the parallelism of the clustered systems. But the question specifically asks about HA clustering (presumably something like the commercial Vertias Cluster Services product we use on our Suns at work). These systems typically have small numbers of systems running the same software with a high speed interconnect to share state on. In the event your main system fails, the clustered backup assumes the role invisibly. That includes taking over IP addresses and having the same open connections, with the same software running in the same state and so forth. (When we test our VCS environments we routinely do so remotely with ssh and the shell we do it from doesn't even see an interruption, although in practice we're suddenly on completely different hardware). So, I'd like to support the first respondants position. I am also not aware of any HA clustering projects for FreeBSD, which as Shaun notes is not to say that there aren't any out there. On a final note, while I support Shaun's idea that the list should be kept clean so that the archive forms a useful knowledge base, I'm not sure I can support the sentiments that we must all be completely correct at all times. The irony of Shaun's apparent misunderstanding of the question makes this even more stark. Have a good one guys, Enno. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200111202341.KAA02627>