just another question that at the moment I can't verify by myself... since I cannot reproduce the needed enviroment for a test.
Let's say we have a number of FAT swing client in a remote LAN, which can connect to a public IP of a remote datacenter in which
an appserver is installed. Let's suppose that we have no problem in opening ports from specific remote IP addresses; when a client
asks for a certain port, these are NATted to the actual appserver.
Now, I know for sure that with HTTP requests the whole thing works. I'm in doubt if it may still work when using EJB Stateful.
I would bet yes...
Has somebody some experience in a scenario like this ?
Well, my doubt was only an hypothesys, there were no evidence that EJBs looked up over internet may not work... I thought that since RMI it's a bit problematic when you bind a remote object on a machine with several addresses and/or behind a firewall, similar problems may occur with ejbs. At the moment, as I stated before, I cannot setup a test enviroment; I'm going to start testing ASAP.
I'm currently call EJBs method incapsulating call in http transport; now I want to test JRMI/IIOP since i think that http calls may lead to performance issues due mainly to network latency. Perhaps a remote Session Facade bean may be more performant.