On 6/15/05, Ersin Er <[EMAIL PROTECTED]> wrote: > Vinod Panicker wrote: > > >On 6/14/05, Ersin Er <[EMAIL PROTECTED]> wrote: > >
--snip-- > > > Is there any performance benchmark for singlethreaded vs. multithreaded > model using MINA? It's there in the roadmap, but no benchmarks are available as yet. Generally speaking, there should not be more threads than are absolutely necessary, since context switches take precious CPU time. NIO raw performance has been found to be not as good as the plain old blocking IO (by myself and others). But NIO definitely provides scalability benefits. Trustin, JMeter has been working on some raw TCP benchmarking module. Maybe we could take a look at that sometime. Regards, Vinod.
