I had been testing JBoss 3 years ago, when setting up a transport market platform with messagedriven beans. It worked well, however, what I was missing was :
Imagine the scenario of a network of JBoss instances running on different hosts interconnected by the internet:
a) I missed means to declare an EJB as distributed - meaning - running on this or that instance of the server, on this or that host, or having data deployed redundantly, defining degrees of redundance, defining propagation methods of updates to these objects, aso. - the whole distributed object handling which I - for now - only know from a product called corso - which is extremely expensive, was missing
b) resource control: I want to be able to specify which EJB is authorized to which resources, and to what amount. E.g: "all of the EJBs modeling workflow between transport service providers should get just about 50% of total cpu resources available, etc."
c) I want to define some extremely cpu-hungry EJBÂ´s as running distributedly, on different machines, at the same time : I do not mean distribution at the instance-grained level (meaning one instance here, one instance on the other server), I mean the possiibility not to cre about WHERE actually the ejb is consuming ejb ressources - but knowing that it is benefiting from the pool of all EJB servers who are registered to _share_ cpu resources for given instances of EJBÂ´s, which - for each instance - run distributedly.
d) distributed transaction handling.
Can I find these features in JBoss ?
If not - where can I find them ?
Are these features non-J2EE compliant ?