Good works Mladen. I found jk a bit faster and it's good to see that we could speed it up a little.
The next step could be to use larger AJP packets (4k too small) On Thu, 17 Feb 2005 14:11:28 +0100, Mladen Turk <[EMAIL PROTECTED]> wrote: > Hi, > > Henri said that he noticed current dev version > of mod_jk being quite faster then previous (1.2.8). > > Although it was not the primary intention to > be faster, I think no one will object :). > So here are some benchmark results from my side: > > JK 1.2.8 single thread > Requests per second: 784.31 [#/sec] (mean) > > JK 1.2.9-dev single thread > Requests per second: 798.01 [#/sec] (mean) > > JK 1.2.9-dev 10 concurrent threads > Requests per second: 918.22 [#/sec] (mean) > > JK 1.2.9-dev 10 concurrent threads with socket_timeout > Requests per second: 910.38 [#/sec] (mean) > > So. Is this a speedup or not ;)? > > Interesting is that new socket_timeout implementation > does not slow down that much. After all it sets the > socket to nonblocking mode before each request, checks if > the socket is still connected and then sets to blocking mode again. > Compared to cping/cpong prepost, the system is almost twice > as faster. Of course it will not detect hanged tomcat, > only if tomcat broke down or some other network problem > happened. > > Cheers, > Mladen. > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]