|
Hello Tim,
> We are using PHP code, which uses cookies, and I need to try and benchmark
> the system. The problem is that each connection will be coming from the same
> IP address and therefore the persistence option in ipvsadm will not work
> because all the requests are sent to one server.
o Why do you want to benchmark a system you're obviously not going to
deploy in any form or shape?
o Can't you disable the persistency option for your test? This is mutually
exclusive to my prior remark and might break your application.
> Is there any method of configuring LVS that might work? Or will I need to
> look for another "HTTP Header" based load balancing system? I'm very
Do you want to have layer7 load balancing (it's the architecture that
dictates it) or can you achive the same results by modifying your php
evaluation requests?
> impressed with the ease of using LVS so far, and would prefer to keep with
> it, but if anyone has any other recommendations, I'd be happy to hear them.
I would need a little bit more information how a typical requests looks
like and what the realservers have to do. A multi tier framework depicted
in ASCII could help too, if you have that much time.
> A solution to the benchmarking problem would also cope with multiple
> requests coming from a company/ISP proxy server as well.
I'm wondering, why you want to have this. It's not a real world benchmark.
You will test things that will in 99% of the case never occur on your
network. YMMV, since I don't know your load balancing topology approach.
Cheers,
Roberto Nibali, ratz
--
mailto: `echo NrOatSz@xxxxxxxxx | sed 's/[NOSPAM]//g'`
|
