Salut,
2008/6/2 Jeanfrancois Arcand <Jeanfrancois.Arcand_at_sun.com>:
> Salut Julien
>
> Julien Vermillard wrote:
>>
>> Hi/Salut,
>> I'm working on benchmarking different Apache MINA version for see te
>> progress made over release (1.1 to 2.0) and I'm trying compare it with
>> latest Grizzly too. I'm using a simple echo server as base, for MINA
>> I'm using the echoserver example provided with the distribution, and
>> for Grizzly the example provided on JF blog. I'm not sure this example
>> is really optimal. So I contact you for some advise about how to
>> configure grizzly. My code is here :
>>
>> http://svn.apache.org/repos/asf/mina/sandbox/jvermillard/grizzly/src/main/java/org/apache/mina/grizzlyecho/EchoServer.java
>> . What I need to add/change ? Have you any rule of thumb for setting
>> mySharedPipeline.setMaxThreads(5); ?
>>
>
> It really depends on the load and how you benchmark it. What do you want to
> do/benchmark exactly? Just throughput, or scalability, or both? Usually the
> value is between 20 and 100 (for the http case at least).
>
My benchmark is simple : a fixed number of connections sending buffer
of 1K to an echo server. I run the client for a fixed time (30seconds)
and I check the total throughput and the CPU load.
It's pretty oriented toward my personal usage ;)
Looking perf against grizzly helped us to improve MINA time for
opening a connection.
>
>> The idea of this benchmark is more for helping our developpement by
>> profilling some use case than bragging on blogs ;) I'm not even sure
>> I'll disclose vs grizzly results outside of dev mailling list.
>
> :-)
>
> A+ et si jamais Grizzly est populaire a Toulouse, j'arrive au plus vite :-)
>
Pour l'instant vu les premier résultat, pour moi c'est mal barré, mais
quand tu veux pour une bierre/cassoulet ;)