I think JPA, which really supplants "entity beans" in EJB3.0 will really be embraced by the JEE5 community.
One problem is simply the delay in getting JEE5 Application Servers out there. So many people are using ORM tools like Hibernate already, that waiting for JEE5 application servers to come out and provide them what they need just isn't practical. And we all know that once a technology takes root, it's tough to get rid of it.
I mean, not too many people are switching away from Windows on the desktop, regardless of how many people might argue that there are better alternatives. Coming late to the dinner table is often the difference between being adopted or not, regardless of the strength or weakness of the technology itself.
I would not say EJB is a failure. I would say Entity Beans may have been a failure. We cover JPA, which I think will have success because it takes into account patterns used by TopLink, Hibernate, JDO, and others.
The book is on the persistence layer, so we do not focus alot on EJB 3. We discuss how using JPA in a managed environment like EJB 3 can make things easier in some scenarios, by allowing automatic propagation of the Persistence Context. I know other frameworks like Spring can give you similar advantages. But in general, I feel EJB 3 is a good thing.
Roland Barcia: IBM Distinguished Engineer, CTO Mobile for Lab Services