Roundtable Transcript, Part 1 (Continued)

Movement Toward SOA
[05:28] Phipps: The place to start out, I think, is to talk about development directions. I'd be interested to hear from around the table what you think is the most significant forward step that Java development has taken this year. I'm making that statement now; I'm going to ask someone to lead off the discussion in a moment, and we'll see how things go from there. Dave Chappell is always the person I turn to as the victim. Dave, what do you think is the most significant development in Java this year?

[06:12] Chappell: I think it's mostly a recognition and movement toward SOA as a recognizable architecture within an enterprise for Java developers.

[06:30] Phipps: You're welcome to expand on that.

[06:33] Chappell: Certainly from my perspective I've seen that a lot of developers who want to use Java want to use it as a way of building new applications, integrating with existing applications, and treating that as a service-oriented architecture and wanting to have a standard way of doing that. So, we've all seen or heard about Web services over the past few years as a way of doing that. But what's been needed is a component model for standardizing that. That's something that within the Java Community Process a number of us here have been actively involved in as a Java business integration effort. That's part of what the second half of the hour's been about.

In the tools area I think we've also seen some interesting developments in Java engineers wanting to think about how to build integration components. Tools that mix the development of interfaces to services with XML technologies like XPath, XQuery, and XSLT for creating services that can manipulate data as its path between applications.

[08:17] Phipps: What do others think about that? Would you agree that those are the key developments?

[08:23] Elloy: If I may add to that, I fully agree. Let me add some perspectives from some of our large customers and prospects [who] I am talking to. First of all, SOA, the higher you go, the more abstract, and the more of a silver bullet it becomes in the organization. Often they think of it as being able to solve world hunger, but clearly that's not the case. The lower you go into the development organization, the more difficult it becomes and that's where the reality bites, which is not only how do you create these things, but once deployed, how do you manage them effectively?

So a lot of the issues that I see are because of the widespread adoption of the vernacular around SOA as opposed to the reality of what SOA solves and the ability to then manage that once deployed. It's clear it's at the very bottom of its adoption at its maturity curve. I think our challenge as tool vendors is to help orchestrate the creation and management of Web services so utility computing becomes more than something CIOs and CEOs discuss as a way to cut development out.

[09:25] Phipps: Do you look upon SOA as the vogue marketing phrase for what was being sold as Web services last year?

[09:31] Elloy: No. I think that SOA is almost at the point of being a disruptive technology. Clearly the Web is mobile and will be when it finally delivers on its promise. SOA in my mind is the uppermost leverage of the Web. It's the most you can get out of the Web from a commercial standpoint, and I don't think we as a world have been successful in doing that just yet. I think we're at the bottom of that.

[09:56] Milinkovich: I think one thing I would add to this [discussion of important developments] is that in addition to SOA and Web Services…I think there's a second. I'll put [SOA] as number one. But I think both BEA and Oracle and some of the stuff they've done with WebLogic Workshop and ADF in terms of trying to simplify J2EE development has been an important step forward the last year. So I'd definitely add that to the list.

[10:21] Renaud: I fully agree. Actually I would perhaps [put that first]. I mean, SOA is a long process. It's something that's gone from buzzword to a lot more concretization this year with model architectures and some definition around certain words, but it's going to go on for a while. I would look at the Apache Beehive project as being probably the most important development in tools this year, to answer your question specifically. If for no other reason than making Java easier to use, which is something we talked about last year and the year before that, has the biggest potential for the Java community at large because bringing in more people into the fold, bringing more people into the club is going to be very fundamental to the success of the Java community at large. Obviously answering the SOA challenge is key, but also opening up the community and getting more people able to be really productive and bringing them to the technology will be absolutely critical.

[11:33] Stover: I think to add on that again, since we seem to be adding as we go around the room…One quick point about SOA, and then I'll move on. Usability or the ease of use of development with Web services and SOA are two different things. Bringing some Web service standards to J2EE…will obviously make that a catalyst for greater adoption, but we have a ways to go. We all know that. So going to the ease of development, [which no one seemed to mention], I've been a big fan of JSF to port more developers who are used to a client/server type of environment where you can build, unfortunately, a functional application that's really difficult to maintain; however, it has to be much easier to build a multitier application, and I think that with JSF and some of the products and the IDEs that are utilizing that technology this year there's been releases that will go a long way as well.

[12:30] Elloy: May I disagree with something?

[12:32] Phipps: Yeah, feel free.

[12:32] Elloy: Just to disagree with something, respectfully, Ben, [with what] you said. From my chair the single most important thing that happened in the Java tools world this year was clearly, and it didn't happen this year, but the propagations happened very aggressively, is Eclipse. So both the rate at which it is being adopted as well as the rate that people are innovating to help make the platform more robust is fundamentally changing the game and the adoption curve of Java in the enterprise.

It's appealing to a much broader set of industries, it's appealing to a different set of programmers at large, and I think that's going to continue. I mean that is a disruptive technology, that's not really a technology. I'm seeing a lot of CEOs and VPs realize (prior to joining Borland I was a VP of app dev, so I've only ever built commercial apps), that Java development is a lot more than the Java developer. So for Java development to be successful you need to treat the act of creating software as a supply chain, which is all it is. As you manufacture any other good, and that's a supply chain to which you have manufacturing, planning, execution, and so on, the act of creating software is really fundamentally no different; it's an engineering process. It's engineering something that happens to be binary and soft at the end of the day. And from what I've seen, a lot of organizations are now picking up on that as a supply chain concept and wanting to get good at it.