On Fri, Jun 20, 2003 at 04:46:31PM +0300, Dusan Djordjevic wrote: > Hi all, > > I am trying to run very loaded (about 6000 connections per second) non > caching www server on Red Hat 9. Hardware is 4 x Xeon 800, 10 gb ram. > Since Squid cannot use all processors (or I do not know how ti make > it). So I need a suggestion, what software to choose for this task. I > tried squid, and I am not satisfied with results. You can run multiple instances of squid on different ports, one instance per processor on your machine. Then use iptables to listen on one port and redirect requests to all instances, thereby having your workload balanced between all squid instances. You can also have an external L4 switch perform load balancing, so it will stop sending requests to a dead squid, though this is more costly of course. Anyway, given the machine you're running this on, you seem to have some resources available fo an L4 switch. Cheers, -- Javier Gostling D. <jgd@xxxxxxxxxxxxxxxxx>
