Community Corner Podcast: Java Champion Adam Bien on JavaEE and Rethinking Best Practices

The very last podcast interview I did at this year's JavaOne was one of the best: my talk with Adam Bien on JavaEE and rethinking best practices. I am primarily a server-side, data center guy, in my software engineering experience. I've been around a while, and have an historical perspective on this. Hence, it was incredibly interesting for me as Adam told me about some surprising (to me) new developments in the enterprise software engineering realm.

Adam's presentation occupied 33 minutes (making it among the longest of our Community Corner 2009 podcasts), but it was definitely time well spent for me. Be sure to download Adam's slides (PDF) so you can follow along as you listen to the podcast.

Adam starts out by addressing the assumptions people have regarding EJBs, for example, that they are heavyweight. This is actually is not true, if you really understand EJBs. They actually have a minimal implementation that is very compact and fast. The full capabilities are extensive, but EJBs can be implemented in concise manner that is indeed highly efficient.

Are EJB applications portable? Extensible? Many think not. But Adam shows that in fact these ideas are illusory, especially when it comes to EJB 3.

What about performance? Adam found a difference of only 3% between a well-designed EJB and an equivalent POJO in his testing. EJBs are actually lightweight in their minimal implementation. But, what they give you is free added capabilities that you don't have to develop and implement on your own - which you'd have to do if you were working on your own trying to extend a legacy POJO into the enterprise realm.

I found Adam's discussion of the conception that "EJBs are too complex" really interesting. I had suggested that the perception of complexity was one reason why developers avoid EJBs. Adam quickly distinguished "essential complexity" and "accidental complexity." Essential complexity is, for example, when you're working with distributed systems. You have to think about things like caching, synchronization. You can't have a genuinely working distributed system that doesn't fully solve these issues. So - do you want to develop your own infrastructure to solve these problems? Well, EJBs do it for you, for free.

What's the simplest possible EJB? One that implements a single POJO. So, why is this valuable? Because implementing the POJO within an EJB instantly gains you all the enterprise-level capabililties that EJBs provide, without any additional work on your part. You just build a jar and deploy it, and suddenly your POJO is available at the enterprise level.

Adam has found that often the perceived complexity of JavaEE and EJBs is really the result of design decisions made by software architects. If the design is overly complex, then the developed EJBs will necessarily be complex. EJBs are indeed powerful. But that potential can sometimes induce a desire to use features of EJBs that aren't really necessary in the specific application at hand -- resulting in unneeded complexity.

I really enjoyed the discussion we got into near the end of the podcast regarding the relationship between stateful components and today's new rich internet clients (RIAs). So-called "fat clients" were in vogue perhaps 10 years ago. Then, suddenly, thin clients (with stateless server-side interactions) were absolutely required. Now, with RIA, we're shifting back to the concept that used to be called "fat clients." Well, EJBs happen to be ideal for addressing this need. Adam tells us how, in the latter minutes of our conversation.

I loved it when, in response to one of my statements about the history of stateful and stateless objects, Adam said:

"In my eyes, everything repeats in 10 years. If you wait long enough, in the next cycle, exactly what you did 10 years before, you get over."

That seems so true to me, thinking about the evolution of client-server applications, the advent of web applications, and today's new RIA focus. We just give it a new name, and everyone thinks it's something that's never been seen before. We agreed that software engineering progresses in a spiral pattern.

Adam has a wonderful knowledge of the history of JavaEE and EJBs. If you're an enterprise Java developer, or if you work in other areas buts you're interested in the high-availability server side realm, I think you'll find listening to my conversation with Adam to be a very worthwhile way to spend 33 minutes of your time. I myself found it fascinating! Check out Adam's site if you'd like more information about his work and his very interesting Enterprise Java ideas.

You can find all this year's java.net Community Corner podcasts as they are published on the JavaOne Community Corner Podcast page.