Interview: Tom Baeyens

What's your motivation to speak at FOSDEM 2007?

My talk will be an attempt to show the value of BPM [Business Process Modelling] and workflow technology to developers. Managers are easily persuaded by BPM technology, because BPM is based on graphical diagrams that managers can understand. Developers are much more sceptical since they see much more alternatives to write this kind of software. Because of the work we have done with JBoss jBPM, we know the trade-offs and alternatives.

We imagine they are positive, but what are your feelings about Sun opening up its JVM ? What do you think will be the impact on the FOSS developer community?

This is the next level of adoption of Java. The OS will loose a bit of its importance in favour of the Java abstraction. The more ubiquitous Java desktop applications become the less difference there will be between Windows, Mac and Linux.

Is the lack of an all-encompassing standard (JSR) for workflow management engines a hurdle ?

Definitely. It's important to know that there is no such standard because there is no consensus whatsoever in the workflow and BPM market. Different approaches, different focus and different environments lead to a completely fragmented market. All vendors see BPM or workflow as a big set of features but in fact, all of these technologies are just use cases of a state machine. Standardization should focus more on the bare bones state machine level versus only on the application level.

Take BPEL, for instance. BPEL is a state machine language to script new web services as a function of other web services. That is a good application of state machines but it is just one. Regrettably, marketing behind these standards has succeeded in coupling this technology to BPM. Since the association between BPM and BPEL is known at CIO level, many BPM projects today fail. Those experiences are starting to ripple through. Reporters and people at CIO level start to realize that BPEL is a good language for service orchestration, but not for managing business processes.

In the next two years, we'll see a shift towards BPMN and XPDL. These standards will increasingly get more focus. Because they serve the goal of BPM much better.

Also in Java, I would really like to see a standard that focuses on the state machine level. The added value of such a standard would be that a lot of developers would get common mindshare about BPM and workflow technology. Just like with relational databases, when people know what it is and when to use it, the technology as a whole becomes a lot more valuable.

What are the major assets of JBoss' JBPM implementation besides being free?

A multi process language strategy. In fact, I think that strategy is even more important to our success then being open source. We have one base technology for implementing state machines. Each process language is just a specific way of describing state machines. So we can support many process languages on top of the same core jBPM technology.

Embededdability. Our engine is really embeddable into every Java environment. You can use it as a standalone product, but you can also take the engine library and embed it into any type of Java application. We have really managed external dependencies very well so that you only need to include libraries that you actually use. Even persistence of process executions is optional so that you can get ultimate performance in case that feature is not needed.

Given the strong position and wide use of open source containers and frameworks of all sorts, combined with the strong standardization around Java, do you think the market proprietary businesses in that area will shrink?

Whether it's open source or not doesn't really matter to most of the consumers. They just look for the products and frameworks that do the trick for them. It's more of a development and business model. Each company has to decide which model works best for them. In recent years, open source has proven itself as a viable strategy to build scalable businesses that can be very disruptive. But building scalable businesses on top of open source software is not trivial to say the least. The dynamics are completely different from traditional closed source development. It's a matter of knowing and leveraging those dynamics in an appropriate way. While the whole software industry is moving towards open source, only a handful of the companies really know the basics of those dynamics. The good thing is that all this focus on open source business will lead to steep learning curves from a business perspective. And over the next 2 to 5 years we'll see these pioneering practices being replaced with proven business models.

Is there a lot of code contribution from non-JBoss/Red Hat employees to the JBoss JBPM project?

Yes. A big contribution is done by the community in QA. They report the vast majority of problems in ease-of-use, documentation and implementation. Secondly, we have a very lively community where people share their insights and help each other to get started with jBPM. We hire from our community. So there is a natural tendency for non-JBoss/Red Hat community members to dry out.

The bulk of the core software is written by JBoss employees. Once and a while people will actually make significant contributions. Also collaboration with other companies like e.g. Bull is now starting to result in joint development of the next generation BPM engine.

What are the current goals for the next release of JBoss JBPM?

The 3.x releases of the jPDL language will focus on more integration and more out-of-the-box features. E.g. jBPM jPDL 3.2 adds email support and improved support for enterprise environments. Also, the BPEL engine GA release based on jBPM is imminent.

For the next generation 4.0, we are working on real pluggability of node types. Which means that it will be child's play to add custom nodes. Also we'll be able to support multiple process languages on top of the same database schema.

And what is the somewhat longer term vision for JBoss JBPM?

I think that the jBPM technology has potential to create mindshare around BPM similar to the relational model in the database world. When someone mentions "This is a relational DB.", instantly you think of tables, columns, primary and foreign keys, join queries and so on. An analogue generic model and mindshare is necessary in the BPM world and I'm convinced that jBPM has got the foundations and potential to realize that goal.

Additional links:

This interview is licensed under a Creative Commons Attribution 2.0 Belgium License.