axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Asankha C. Perera" <asan...@apache.org>
Subject Re: Axis2/Rampart WS-Security performance
Date Fri, 12 Feb 2010 04:25:27 GMT
Hi Amila
>
>     > In the very first scenario upto 640 concurrency level both have
>     same through put while for 1280 threads UE has a sudden increment
>     and for 2560 other ESB has a relatively high change. What could be
>     the reason for that?
>     Which set are you referring to - the WS-Security case? Sorry I do
>     not see what you mean..
>
> 1. Direct Proxy Scenario 500b case.
> And also
> 1. Direct Proxy Scenario 1K case
> for 640 concurrency other ESB has performed well but 1280 and 2560
> done well.
>
> I am just wondering reason for these anomalities.
I think the anomaly is most possibly due to a full GC of the JVM - but
that its limited to the 500 byte / 1280 users case only - for the other
ESB. The usual trend would be for the TPS to increase gradually, and
then drop again after the sweet point of the ESB for the load considered.

If the same test is repeated multiple times, and the averages taken, the
results could eliminate these spikes - but the process could take some
number of hours more. These results were obtained after a warm up run of
1000 iterations at 20 users for each of the 4 scenarios - as we did in
Rounds 1, 2, and 3 before. This was to trigger the JVM to compile the
code with optimization - usually reached after around 10,000 iterations
of a method.

cheers
asankha

-- 
Asankha C. Perera
AdroitLogic, http://adroitlogic.org

http://esbmagic.blogspot.com





Mime
View raw message