Terracotta Discussion Forums

My app is designed special for making a benchmark tests. I would like to test your system and compare with other open source solution for clustering (jboss as, jessica2).
As you saw there is only operating write, but I thought that fine-grained replication makes that terracotta will be better than jboss as cluster with coarsie-grained replication level (I configure, that whole session object is replicated) . So, is there posibility that my results are good?

With best regards

Can you post the code for SharedObject then?

Here is SharedObject class...

I am sorry we let this thread get out of date - Java One is this week so we've been pretty busy.

It appears the inputs take parameters - what are the parameters you are passing from your test?

Of course I attached some parameters to request. There are "size", which determines what is the size of SharedObject, "replace", which means how many cells array in SharedObject has (numer of cells is "size" divide "replace", e.g. 1024/256=4 cells ). "replace" is also number of bytes which are changing during invoke SharedObject.replace() method. The thirh parameter is "manage" and it's only option for JBoss AS cluster. Terracotta doesn't have to call setAttribute method, but JBoss has to. When I run this servlet with Terracotta the if condition with checking "manage" parameter is never true.

I hope this give you "new light" to this experiment.