Ideas of test criterias :
Make a very simple HTTP server and try on it, the dozen of HTTP tests existing specialy the ones made for Apache HTTPD =)

Besides stateless protocols like HTTP, we'll have to test some stateful ones like ping pong.

Postponed to 3.0

Hi, recently I've been asked to do some research on MINA performance. I think that performance could be measured in different layers. Performance of particular family of Acceptor and Connector pair. Performance of filters with dummy transport. Performance of codecs. Some complex protocol stack performance (i.e. http over tcp). I am interested to invest my time into the benchmarking code. So if there are some benchmarking requirements or ideas in MINA community, I will appreciate a chance to know them.

Hi,

this is a very interesting proposal ! We are really interested to get the result you will find, of course.

Currently, we have no formal requirements, except that it would be great to be able to run those benchmarks on many different platforms/JVM/OS. (I was thinking about W$/Linux/Solaris, Java 5, 6 or 7, from Sun, IBM or Bea (jrockit), mono proc/ multi proc... Typically, we have a 8 way CPU available, with 9 injectors, on a 1 Gb ethernet, with all the mentionned JVM, so if the benchmark could be run on such a configuration, that would be just perfect.

Otherwise, I suggest that you post to the DEV mailing list your proposal, questions, results using a tag like [MINA bench].

This is an exciting proposal

Its an interesting proposal. What most people like to see is Hardware configuration + MINA's processing/handling power. I could help out with MAC benchmark (if needed)

Almost 10 years old proposal, which never came to realization. Killed it.