[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On Sunday 21 October 2007 21:49, Richard Brown wrote: > I have 20 off p3's 700mhz. I assumed that if we linked 10 together in > a cluster we would have a super-computer that we could run a network > off. i.e. the cluster would be a server to lend power to the other 10 > in the network. I didn't fully appreciate what a cluster was. > > So the first question is what is a cluster? What is the point to it > and why isn't it a server like I wanted? The link Rob provided in the follow up e-mail gives a reasonable description of clustering options. While we were on the topic I was wondering if anyone knows anything about asymmetric clustering? Symmetric clustering involves collections of more or less identical machines. I've been trying to find out if anyone is working on asymmetric clustering with no real success. The idea being as follows: In an organisation it would seem sensible to provide as near 24/7/366 cover where possible. High availability clustering would do this until its upgrade time. The current setups seem to require a whole batch of near identical machines - you can add more to the current cluster but after a couple of years you cant normally find the same product so you have to upgrade. I'd like to just be able to get in a new machine, stick it on the network, lob in a CD and type in a couple of names and passwords and have this machine become a useful part of the cluster. There seems to be some research in this area but I've not found a distribution that attempts to achieve it. I'd love to give it a go - set up a hole bunch of different machines for a HA cluster and then shoot one to show it still works - might make great UTUBE! Tom te tom te tom -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html