IBM"s Rumbaugh scorns Microsoft UML stance

24.03.2005
Von 
Paul Krill ist Redakteur unserer US-Schwesterpublikation InfoWorld.

Jim Rumbaugh is an IBM Distinguished Engineer who currently leads software modeling efforts at IBM Rational. Along with Grady Booch and Ivar Jacobson, he was part of the "Three Amigos," a trio responsible for developing Unified Modeling Language, which was adopted by the Object Management Group in 1997. He also has participated in development of the Rational Unified Process and was chief developer of the Object Modeling Technique for object-oriented analysis and design. InfoWorld Editor at Large Paul Krill met with Rumbaugh at the SD West 2005 conference in Santa Clara, Calif., last week to discuss UML, SOA (service-oriented architectures) and ESB (enterprise service bus) technology. Rumbaugh had some harsh words for Microsoft and its fence-sitting on UML.

IW: Could you talk about the background of UML and how you were involved in that?

JR: Yes. (There were) a bunch of methodologists and each had their own methods (for modeling). And the people kept asking us, Well, why don"t you unify all of it? And all these people were at different companies (so) there was no incentive to do that. The way UML came about was Mike Devlin, the CEO of Rational, hired me. Ivar Jacobson and Grady Booch (also were) at Rational. Now there were three of us, but one company, and that was suddenly critical mass, and Grady and I first put together what we had called the Unified Method. And actually Ivar suggested calling it the Unified Modeling Language. Now at that point, there was incentive for other people to jump on board. That"s when the whole (effort) got started.

IW: What was the goal?

JR: Well, the intention was to get some unity out of all these different modeling approaches, and it was to standardize them on one approach that everybody would use. And it eventually resulted in UML. Now a lot of other people besides the three of us and besides Rational eventually got involved, and that"s what made it successful. And now, you know, UML has pretty much driven all of the other methods away. Even the practitioners of other methods eventually gave in and started using UML. So I think UML has succeeded.

IW: Microsoft is not really supporting UML. They"ve told me they don"t see that much demand for it. What kind of problems does that present?

JR: We"ll see if they"re really on board or not. They"re supporting it in some ways, too. They seem to cover all bets. I bet if they find (that) enough people want it, they"ll come around eventually, if they can"t beat people into doing it their way. But again, a lot of people seem to think UML is useful. And just because Microsoft says something else, (that) doesn"t mean it"s not going to happen. They haven"t managed to have it their way in every possible area, that"s for sure. (Editor"s Note: Microsoft did provide support for UML 1.4, but is not backing the newer version, UML 2.0. Borland Software, however, has unveiled a UML 2.0 designer to function with Microsoft"s Visual Studio tool.)

IW: You talked about the importance of model-driven architecture. Is UML a form of that? What are some of the MDA-based technologies?

JR: UML is a modeling language, so it"s a way of expressing models. MDA is a way of going about and constructing models. So the two are sort of orthogonal. You need a way to express a model and then you need a way to develop it. So one is a development process and the other is, in the sense, the language, the format, whatever you call it -- the representation.

IW: In your presentation at the conference, you talked about SOA and standards. You never mentioned Web services. Why was that omitted?

JR: Well, Web services are kind of SOA, so in fact it all falls under that title. Web services were sort of the first example of SOA, or a fine example anyway.

IW: You mentioned ESB. Does IBM have a nEnterpris eService Bus product or are you going to have one?

JR: It"s essentially embedded in our (WebSphere) middleware. It"s not really a separate part. In a sense (ESB is) a concept that the products need to support. So it"s not something you separate out. It"s a property that your middleware needs to have. It"s a term for this property that you need to easily be able to call anything from anything else. And not let things like machine formats, languages and platforms get in your way.

IW: So what is ESB based on? It"s pretty much a messaging bus, wouldn"t you say?

JR: Well it"s based on the idea of exchanging clear messages between different things. So it"s basically sending messages back and forth (between) applications.

IW: How prevalent are SOAs at this point?

JR: Yes, as we pointed out before, Web services are a kind of SOA. So (users are) certainly doing it and we expect to (see) it to a greater and greater degree. And part of our job is to make it easier to happen. The problems up until now have been the barriers. Things like the difficulty of connecting across platforms. So IBM is (building) middleware. The job of middleware is to enable people to construct applications that can communicate and can do the things they need to do. (There"s) already plenty of it out there, and I expect it to expand a lot.

IW: So SOAs are based pretty much on Web services at this point?

JR: Well, that"s certainly a major part of it. But there"s certainly plenty of SOAs that wouldn"t be necessarily Web services. We see the future of SOA as taking these monolithic systems and chopping them up into essentially bite-size pieces that can be easily reconfigured and swapped in and out.

IW: Will a company moving from a monolithic architecture to an SOA have to spend a lot of money? If I had somebody from Sun Microsystems in here, he would probably say IBM would sell IBM Global Services offerings for SOA deployments. Is it going to cost a lot of money to switch to a service-oriented architecture?

JR: Well, you don"t do it all at once. One of the messages was that this is something you do incrementally. The way people and companies get burned is when some guy from the top says, "We"re going to convert everything to new Great Idea X, and we"re going to do it overnight and everybody"s going to have to go along." Those are the big horrendous disasters you"ve seen where it just doesn"t work. The whole idea was you do it one piece at a time, so you never get a big hit. You do it on the areas where it pays off.

What it may take at first is a change in attitude. People are used to the idea (of having) this tight, closed shop in the IT house. Well, the idea of opening it up and contracting out some services somewhere else could be a real culture shock. That could actually be the real impediment to getting started, these cultural attitudes and political attitudes inside companies. People worry about losing turf and things like that, and that"s usually the problem with businesses is they get in the way -- and then that"s the CEO"s job and CFO"s and CTO"s job, to get past those barriers.

IW: You mentioned choreography. What"s your impression of BPEL?

JR: Well, we"re using it. It"s a fairly low-level choreography, but it"s important to work on top of standards. So we have products that work on top of that.

IW: In the Java vs. .Net war, so to speak, is there going to be a winner?

JR: I suspect that we"re not going to have a clear victor where the other side just gives up. I mean, I think they"re both going to survive.

IW: Although Microsoft seems to have found more of a way to make money off .Net than Sun has off of Java.

JR: IBM is also behind Java, too. Let"s put it this way: I think Microsoft has a formidable opponent there.

IW: Do you think IBM is the lead with Java technology, even ahead of Sun at this point?

JR: Well, I feel that IBM is pushing it strongly, and I don"t think we"re expecting Sun to stop supporting it. But I think certainly IBM is one of the major leaders anyway.

IW: How has the merger between IBM and Rational worked out? Does Rational feel constricted in any way?

JR: Actually I think a lot of people at Rational feel that the Rational products have a much bigger scope now than they used to, because IBM has a much wider scope than Rational, and we"re selling products where we never would get in before. So, in fact, we"ve actually seen increase in scope.