You've had some harsh words for the Service Component Architecture (SCA) specification. What are your exact concerns
about it? From a technical standpoint, SOA is about interoperability and open standards. The reality of business from the proprietary competition we face in SOA is that they make a lot of money in those spheres -- the integration, the distributed transactions. All of that is where WebLogic and Tuxedo at BEA [Systems], for example, makes money, and BEA leans solely on that. IBM derives a lot of revenue from its integration and its distributed transactions, CICS and middleware. The vendors don't really want rapid commoditization by open source players like JBoss. They don't want to see that happening to SOA. They're not afraid of an endgame, specifically IBM, in 10 years, where it's all open source and we all live on services and maintenance. They're afraid of the rate at which we transition to that world. Sun's announcement with SeeBeyond really represents an acceleration of the timeline.
Back to the standards, I think what's going on is why were the Java bindings done for SCA, which is an attempt at a closed standard? Why wasn't it done through the JCP [Java Community Process]? Why are these folks pushing their implementations and then seeking the standards approval? Because they want to milk it as long as they can. It's about a business decision. I'm not knocking them. I would probably do the same if I were in their shoes.
The message to the end-user community should be very clear: In 2006 what's shaking up again is the dichotomy between a group of vendors hyping up a marketing message to push licensed technology and give it a standards feel, even though it's a closed standard -- but people should not be mistaken about this. If you decide to go that route, as an IT user, you're basically choosing closed source and closed standards, which may be your strategy.
I think that's a niche market, though. The other side of the dichotomy is this volume distribution of open standards based technology with open source. The IT end-user organizations should think long and hard about the value they get from proprietary vendors with closed source standards and the risk they expose themselves to by continuing down an SOA path with a license-based business model.Just playing devil's advocate, are Java and the JCP up to the challenge of SOA or is that a level of abstraction beyond it?
The short answer to your question is yes. In fact the JCP is exactly the kind of venue you need if you're going to deliver on the truly interoperable world of SOA. The Web services standards are one way, but they may be a step back from a technology standpoint. The way they do the communication for Web services is just not effective. It's been a technological slowdown and barrier to entry to this new SOA world. So the JCP to me is absolutely the place this should happen. Otherwise, you're breaking the promise of portability and interoperability, which is what SOA is all about.
That being said, anytime you have design by committee, that comes with its own inertia. Inertia is natural and it can be frustrating. For the reasons we've just talked about, some of the vendors prefer to fight a political battle, to splinter the standards so they can move at their own proprietary pace, breaking the SOA promise of interoperability, but wrapping themselves in 'Oh, we're building to standards now.' That's doing a disservice to the end-user community.
The truth is the end-user community, if you read the latest Gartner reports, is barely starting to adopt SOA. This is the fundamental disconnect I was referring to, which is the vendors are hyping the story so you buy everything. We operate at JBoss very differently. Since we don't have licenses, we have no interest in running ahead of ourselves. We know what people are doing. They're adopting SOA piecemeal. We need to make sure the story we deliver is one of interoperability between all vendors and that means one thing: true open standards. The JCP has its set of, I wouldn't call them problems, I would say annoyances, such as its speed and the politics to go anytime you do something by committee, but that's the way it's got to be, by definition.Along those lines, JBoss obviously has its roots in Java, but is that something you'll grow beyond in the coming years, not abandon it per se, but expand into new arenas?
Middleware is a big place. It may seem like a niche from the outside, but from the inside it's a big niche. There are other areas we're looking at, very specifically PHP and .NET. You've seen us collaborate with Microsoft, announcing another partnership. It was more than marketing noise. You're going to see some results coming out of us and Microsoft in 2006, something very well-advanced and in the pipe. You're going to see some amazing technology of integration of PHP, Java and .NET coming out and we're very excited about that. Any plans to get into the Ajax business, on the portal side for instance?
Absolutely, very probably in 2006. When companies make mistakes on the road to SOA over the next year, what mistakes will those be?
End-user IT organizations are not dumb. They're sort of past the vendor hype after the indigestion of the [dot-com] bubble. People are very critical of technology and the value it brings to the organization. The biggest threat was people would think SOA is junk and not do it, but I see people doing SOA the right way, which is to get top-down buy-in. SOA's about coordinating different divisions, that's the most difficult thing to do in corporations. SOA will fail if you don't coordinate. A little bit of top-down buy in enables you to do your pilot projects and take baby steps and eat the beast one piece at a time. Earlier, you mentioned your own personal paranoia, so what is it that makes you paranoid?
We're a 150-person company and our business model is highly disruptive to the market. We have heard from ex-IBMers that have left that they speak about us in executive circles, citing the name of Microsoft, that it's Microsoft all over again with a volume player -- not just us, but MySQL as well -- stealing the thunder of and competing very favorably against DB2 and WebSphere. What I mean by disruptive is that if you look at our profit and loss sheets and how we operate -- and I know this will sound delusional, but trust me it's paranoia -- the fight is not even in our advantage. Here I am being Mighty Mouse with 150 people and knowing that I have an unfair advantage toward the competition.
And the paranoia that we're so disruptive yet so, so small, we have to think like cockroaches here. There's a big war with Sun, with IBM, with everybody. We cannot afford to not be deeply paranoid as an organization. We exist because we're paranoid and we have to be going forward. Paranoia can be negative or positive. Paranoia being negative means you've become crazy, that you've become so paranoid of the world you don't see the reality out there.
At JBoss I'm trying to turn the paranoia into a positive outcome. The positive outcome is this: Guys, we are not geniuses. We are good, but so is everybody. Never underestimate the competition. The positive result would be a constantly learning and adaptive organization. JBoss today is different from JBoss three months ago, is different from JBoss six, 12, 24 months ago. We will be successful at JBoss only if our paranoia results in humility. We've got to be adaptive. We've got to be agile. That is only done if we are humble enough, and yet arrogant enough, to know we can pull it off if we learn how to respond … and that has got to be rooted in paranoia.
Today, we're doing great. Tomorrow, it may be a different game. We don't have the luxury of the big balance sheet like all of our competition does.