Interview: JBoss CEO Marc Fleury

Now that JBoss Inc., the stewards of the wildly popular JBoss Application Server and other open-source middleware, is on its way to fully license J2EE compatibility, some see the possibility of JBoss disrupting the commercial J2EE app server business the way Linux has disrupted the proprietary Unix business.

JBoss CEO and founder Marc Fleury recently spoke with CRN Senior Editor Elizabeth Montalbano about why he's committed to open source as a lucrative business model and how things have changed between his company and Java steward Sun Microsystems since JBoss Inc. became an official J2EE licensee. Fleury also took a firm stand on why, despite objections from IBM and open-source proponents, Sun should continue to oversee Java licensing and compatibility.

CRN: Could you talk about why you don't think giving Java to the open-source community is a good idea?

Fleury: It may seem ironic on the surface of things, but really I hope to make the point clear to you. Essentially, [the question is] what would the Java community at large, users etc., gain by open-sourcing Java, the virtual machine itself? The state of the virtual machine is actually very good. The portability of Java is not to be proven, it has been proven, and that is what made Java. Why is Java today the dominant corporate development language? It's because the portability of Java on the server side amongst Windows, Solaris and Linux is superb. That works.

Would open-sourcing that--and the loss of control from Sun that it entails--work? I don't know. Obviously, they're talking about, 'Oh, if we do open source, then Sun would still control the compatibility." I still have to see a net positive value, and it hasn't been articulated by any one of the proponents in any convincing fashion. [Open-source activist Eric] Raymond's letter [calling for open-source Java] was, and I'm sorry to say it, a little bit ignorant. It was emotional. It was bordering--which is surprising for Raymond because he's usually not in that category--it was bordering [on] open-source zealot and open source for the sake of open source. And Raymond has accustomed us to [his having] a lot more pragmatism and a capitalist open-source approach. This was almost a rant. ... Frankly, I'm part of the open-source community and I was a little bit embarrassed by the lack of knowledge that Raymond had about the state of Java and enterprise IT.CRN: What about IBM's letter from Rod Smith calling for Sun to open-source Java that followed Raymond's' letter?

Fleury: The thing on IBM was, I couldn't shake the feeling that it was a political move. ... I thought it was political [backing] of an emotional and ignorant letter to purely [advance] corporate goals, which is IBM would like to be the controller of Java. I think it was just an emotional point. I haven't read an argument that this would be good for everyone involved in any convincing way. The only thing that could evolve better in the Java virtual machine world is, yeah, we would have an open-source implementation now. The specs would still remain with the JCP and Sun, and the JCP so far is not that bad. It's pretty good and pretty open. The living proof is that guys like us are writing [the Enterprise Javabeans] EJB 3.0 [specification] with Sun. As far as the industry goes, it's very open, and specifically for the virtual machine layers, I don't see that it's broken.

We have fought for open-source implementations. We have never fought for taking the standards process out into the open-source [community] because I'm not sure it would work. Coming from the professional open-source community, we see the value of communities, of large groups of people collaborating. And certainly this is present in the JCP. Also, we see the value of one company being responsible at the end of the day. It's a very reassuring message of the benevolent dictator, and Sun certainly has an excellent track record as the benevolent dictator of Java for the industry. They have put together a multi-billion dollar industry with their partners. The value of that mix of a control point with large communities is definitely a dynamic at play with open source, and we are advocates of that in professional open source, which is not this [idea] of nobody is responsible in open source--because that only serves IBM, because they are the biggest [player].

CRN: It's funny to hear you say these things because only about a year ago JBoss and Sun were having a serious public fight over J2EE licensing, and you said some negative things about Sun.

Fleury: It's true, and you know ... some people talk about whiplash, and things like that. Definitely on the surface of things we have changed our tune. But what we were fighting for, which we got, was the rights to license and [develop] an open source implementation [of J2EE], which was not available to us before. It was just not possible. So Sun delivered on that front, and then I don't have a problem with the rest. I'm not calling for everything open source ... This is a real industry. I'm not calling for status quo, but in fact, there's no status quo. [The Java community] is moving well, our participation as federated and professionalized open-source communities within the JCP is a big sign of health and strength of that community. So far, so good. Definitely there is a friendly message between Sun and us, but it's not like we've changed our views on things. We've always called for an open sourcing of the implementation. We got that, and we're done.CRN: How is your relationship in the commercial application-server community? Do you see your competitors being nervous about your presence?

Fleury: I think there was a lot of nervousness about open source because our model, and I mean business models now, are disruptive to the industry. There was also a lot of ignorance about open source. I think what Linux is proving, and MySQL is proving, is that open-source vendors now, at the end of the day, are normal players, in fact behaved players, in an industry. And an industry doesn't evaporate overnight because you have an open-source player.

I'm really glad now that Sun has turned around with respect to open source. Now that we've paid our dues in licensing fees, we are a participant in that community. What we see from a lot of Sun employees is a warm embrace. ...

It's a maturity step for the whole industry, figuring out our business model, our company model and how we play with the others--and for the other companies to understand that open source is there, it's not going away, and [asking], how do we work with these guys? Good things are coming out of the collaboration. It's just a normal maturation of all the industry players.

CRN: What is your take on Apache and the opinion that some have that since Apache runs a tight ship and has a model in place where they ensure compatibility with their implementation, that that might be a good place for Java to reside?Fleury: Does IBM say that?

CRN: No, they didn't say that to me. Some developers and solution providers that build applications have mentioned this to me that Apache might be a good steward of Java.

Fleury: I'm not so sure. I'm talking about the C-code base [group that built] the original Apache Web Server. [The Web server] is a big success, a big open-source success. The group, though, it's not a group of professional [open source players]. The nonprofit nature really bothers me actually.

CRN: Why's that?

Fleury: Because if there is no one for-profit entity, then the biggest guy wins. And right now that's IBM at Apache. If you look at what happened with [the] Axis [open-source project], it's a proof point. But please don't mischaracterize my intentions. I have a lot of respect for the Apache Software Foundation and the C-code base there, and the same thing goes for IBM. IBM has done a lot of good for the open-source community at large.Now that I've given you my disclaimers, look at Axis. Axis was the SOAP reference implementation, a very important stack, given to the open-source community at Apache. A year ago, [I thought], it's Apache, it's [from] IBM, it's safe. Let me go there, let me adopt Axis, and we have based some of our work on Axis. In that respect, that decision was made just like any other end-user IT organization would make that decision. We said it's safe, let's go. Then what happened is the project took off, people started contributing, IBM contributed a lot and when it was stable and almost a product, then IBM forked back into IBM Corp. a proprietary version, which is allowed by the BSD-style license that the Apache Software Foundation has. That led me, my guys, our group, and all the other users that contributed to Axis in a void.

CRN: But wasn't the code IBM worked on already part of the open-source community? Couldn't you use that?

Fleury: There is a branch that is still open source and is still the Axis branch, but essentially, all the development is gone because IBM was the biggest one, and they have gone back proprietary. [So IBM said], "Thank you, our corporate decision says that it's more advantageous to us to take it back to us proprietary, see you." Then there's no commitment [to open source].

That is very disturbing to me because all of a sudden what I thought was a safe bet as a user of that technology, not even as a developer, is gone. And now I'm scrambling to save that, and how do we do it, and maybe we start committing ourselves to Apache. I can't imagine how the developers that committed to that project feel about this because all of a sudden that product roadmap for that project disappears because the biggest committer to that project has gone back into the proprietary world.

CRN: Eventually, SOAP as a standard did go back out to the community, though.Fleury: I'm talking about the implementation. The implementation is gone. So the rug was pulled out from under our feet, basically. So I question the licensing model there, which is middleware-vendor friendly. And there are very few of us [middleware vendors]--there's JBoss, BEA and IBM. It's quite friendly to us. Essentially, I could do the same to Axis, because that's friendly to me, as a vendor. But how is that friendly to end-users and end-users of that project, and how is that friendly to developers? It's wrong in my mind.

CRN: You're afraid that might happen with Java if it were open source.

Fleury: The bigger guy could just come in, steal your wares and run. That's the problem with the licensing and the fact that there is no company. Apache, it's a Web site. It's a nonprofit organization. You're not doing deals with someone; you're doing deals with a Web site, to some degree. There is the Apache Software Foundation as a nonprofit and I'm really cool with that. That's a success. But is it the right structure for something as critical for infrastructure? I simply don't know. What Apache does well is a lot of little projects, utility tools, definitely the parcers are great, Tomcat. But we are the leading sponsors of Tomcat today. ... To sustain that model means commitment from us. It's not self-sustained. It depends on contributions and charity. I don't think it's the appropriate model for professional open source.

CRN:: So ultimately, you think there has to be a vested financial interest in open source for it to be successful in the IT industry.

Fleury: I want to answer this very specifically. There are many models of open source. There's the nonprofit open source and Apache is a success in that respect, in the nonprofit sphere. But then Linux [vendors] MySQL and JBoss are for-profit, and we believe strongly in the professional open-source model, this new generation of companies--emphasis on companies--that have not just embraced open source, but were born out of open source. We claim we are the future of open source, in a sense.For-profit open source is a good thing for open source. For-profit open source is a good thing for the developers involved in open source. Everyone here has stock in this company. And if one day there is liquidity, all my guys will make money. And that's a big motivation for a lot of people. The image that open-source developers are necessarily students with ponytails on the communistic fringe of things is a misrepresentation, specifically in the open-source Java camp, where we are all professionals with advanced degrees from very big universities and we could be making very big salaries in the private sector.

I'm very proud of the fact that we normally match but in certain cases exceed what somebody can make and then on top of it we have stock, and I think that motivation is a healthy thing. I'm somewhat capitalistic in that sense. In between the greed of normal capitalism and the communistic fringe, there is a pragmatic middle road that says look, open-source is a category, it's a profession, we do it seriously, it's our life, it's our passion, we're making a good living at it and blah, blah, blah. To me, that's a very healthy message that you hear consistently from the second-generation open-source companies like MySQL and us.