Gonçalo Jesus wrote:
Well...you're partially right about the Heap Size! I mean partially
because that is one of the parameters of the test, that definitely would
influence the number of requests executed and the time the server could
hold on. These tests with Axis2 are included in a batch of tests that
try to comprehend the most known platforms for Web Services
(Axis+Tomcat, Axis+Tomcat, JBoss, .NET, gSOAP, Axis2+Tomcat,
Axis-C++/Apache,...). I hope to show you guys the results in a near future!
But regarding Axis2:
Initially i thought that the low quantity of requests executed was due
to the low Heap Size i gave it (first test it was with heap size to
128MB). But when i increased the Heap Size to 1GB the results were
pretty much the same (just about 100 requests more...)
You can see now why i send the mail asking for help! I thought that was
something i missed...
About the profiling...now i don't have much time, the deadline for the
paper i'm working it's coming really close! But if you want to really
test the Axis2 platform...if you could deliver one release that doesn't
have this problem in the next two weeks, i'm going to have access to a
400 machine cluster just for these kinds of tests...
Thanks,
Gonçalo
Ajith Ranabahu wrote:
1. I'd like to see a draft of the paper; email it to me and I'll gladly
comment.
2. Which cluster are you using? PlanetLab? One of the EU Grids?
-steve
