Yes, you read correctly. OpenEJB can be embedded and treated as your very own personal EJB container.

If they can have Local and Remote EJB’s, why not Local and Remote EJB Servers too?

Haven’t you ever wanted EJBs without the heavy? I mean you need the "heavy" eventually, but not while you’re developing. Well, there’s the advantage of an EJB implementation that was designed with a very clean and well defined server-container contract, you can cut the server part out completely!

So, if you wish to access ejbs locally and not in client/server mode, you can do so by embedding OpenEJB as a library and accessing ejbs through OpenEJB’s built-in IntraVM (Local) Server. Why would someone want to do this? * Your application is a server or other middleware * You want to write an app that can be both stand alone and distributed * To test your EJBs with JUnit and don’t want to start/stop servers and other nonsense * Imagine the power from being able to use your IDE debugger to step from your Client all the way into your EJB and back with no remote debugging voodoo.

In this case, your application, test suite, IDE, or client accesses beans as you would from any other EJB Server. The EJB Server just happens to be running in the same virtual machine as your application. This EJB Server is thusly called the IntraVM Server, and, for all intense purposes, your application an IntraVM Client.

There are some interesting differences though. The IntraVM Server isn’t a heavyweight server as one normally associates with EJB. It doesn’t open connections, launch threads for processing requests, introduce complex classloading heirarchies, or any of those "heavy" kind of things. All it does is dish out proxies to your app that can be used to shoot calls right into the EJB Container. Very light, very fast, very easy for testing, debugging, developing, etc.

!http://www.openejb.org/images/diagram-local-server.gif|valign=top, align=right, hspace=15! \{include:OPENEJBx30:Embedding}