Salut,
Niall Gallagher wrote:
> Hi,
>
> First of all, great project. I have been watching with interest for a long time now. I have always downloaded the latest, and with 1.9.x I have found Grizzly in a very stable state. However, I am a little ignorant to the strategies Grizzly uses under load. I know they exist, including changing from persistent connections to connection close under load. Given this I have a question. I myself write an open source NIO server, and have paid particular attention to performance. With 1.9.x I have been able to sustain a large test on Grizzly comparing it to Simple (http://www.simpleframework.org/).
>
> In my performance tests I am particularly interested in latency. I have found Grizzly has really great performance during the initial stages of the test. However, although there is very little latency deviation during the test and throughput is maintained it seems to slow up a bit after about 10,000 requests. I would assume due to strategy change??
>
> For example, here are two runs I have made. Grizzly shows exceptional performance up to about 10,000, but looses out badly to Simple in a sustained test of 1,000,000. It comes out 50% or so slower. Perhaps it peaks early. Before I publish this I want to ensure the tests are fair? Is there something I need to do to tune the performance? Here is what I see.
>
> Throughput:
> http://www.simpleframework.org/performance/grizzly/run1/ScalabilityApacheBench.png
> http://www.simpleframework.org/performance/grizzly/run2/ScalabilityApacheBench.png
>
> Latency:
> http://www.simpleframework.org/performance/grizzly/run1/LatencyApacheBench.png
> http://www.simpleframework.org/performance/grizzly/run2/LatencyApacheBench.png
I have to admit ab benchmark are not something I usually like:
http://weblogs.java.net/blog/sdo/archive/2007/05/how_to_test_con.html
Also you just set the WorkerThread to 5 :-) How many thread are you
configuring with your framework? I would also like to see the server and
ab logs to see what is the issue. Which JDK/OS are you running on? What
kind of network. And more important, what you framework offer in terms
of http :-)
>
> Criteria:
> http://www.simpleframework.org/performance/grizzly/README
>
> Do I need to tune the server somehow?
Well most probably yes :-) But your goal is to claims your framework is
faster than Grizzly, right? A lot of framework did that as well, but at
the end it is always quite easy to run something different (like Faban
above) and demonstrate the contrary :-) Anyway, I'm interested helping
and see how it goes :-)
A+
-- Jeanfrancois
>
> Thanks,
> Niall
>
>
>
>
>
>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_grizzly.dev.java.net
> For additional commands, e-mail: dev-help_at_grizzly.dev.java.net
>
