BEA, SAP and some of the other Java vendors have been critical of the extent to which Sun controls the Java Community Process, what is the IBM position on that?
I don't know that it's gotten any worse and I also don't know that it's gotten any better. It's been working for a number of years. Sun still has the poll position in the process. We all respect what got us to where we are, so we're not going to be disrespectful to the Java Community Process. It served us all very well, including my colleagues at BEA and SAP. It has a purpose. We know where its strength and weaknesses are. We need to be careful not to let it slip away and cease to be the place where we (vendors) come together. But IBM actually has diversified. There are other forums where we are standardizing. We're not solely in the JCP. There's Eclipse. There's OSGi. There's Apache. We have our eggs in multiple baskets. So there's more than one way to build an SOA?
SOA is an infrastructure. The power is that we all agree. We did a terrible job as an industry in the past of agreeing on how to build an app. We did a terrible job when you compare it to the hardware industry, the telephone industry, the electronics industry. Now, projects built around something like OSGi or Spring with componentization and modularity mean we pop things in and out of either the WebLogic server or WebSphere and get much more reuse across platforms. Java EE 5 how important is that to WebSphere?
It's very important. I wish EE 5 was EE 4. We would have had it three years ago. There are things in EJB 3 that are way cool. It's a day late, but it's here. I was just in a room with some of our most religious customers around WebSphere. They vote with their dollars. They asked a question about EE 5. We asked them the question you're asking me and 60 percent of them put their hands up. So yeah, Java EE is a very important part of WebSphere now, but it's not the only game in town.
That's one interesting revelation if you're open minded to other server profiles beyond Java EE. Look at supporting Tomcat or PHP. These are profiles for servers. They are very interesting. There's the LAMP stack. We've looked at that. There's some interesting lessons to be learned there around productivity and how to build Web-oriented applications. So Java EE is very important. It's paid my paycheck for many years, but there are other games out there being played. We want to keep playing the Java EE game, but we want to keep playing some of these other games too. We want to have WebSphere Community Edition with a server profile that's just Tomcat and some management software with WebSphere underneath.
So one of the things I've been reflecting on is what happens if we embrace the Web more as the platform? We built application servers and relinquished the role of the Web servers to a lighter transport role. What if we made the Web server truly the application server? It's a popular term now, REST. REST has always been around. So what happens if we really built around REST with URLs and GET, POST, PUT, DELETE as the four operations on those URLs. How far can you go?
You think about lightweight environments. One of the things I think about having done in WebSphere is can we build a radically simple environment on the Web for building Web-oriented applications? I think the answer is yes. You can do WOA (Web-oriented architecture), the agile cousin of SOA. It's doing SOA on the Web, Web SOA if you will. So one of the things I'm driving very hard at with my incubation team is experimenting with a REST server. Are we going to see results from REST soon?
One of the things that can be done immediately is we can put REST interfaces on some of our existing enterprise services to make them easier and more affordable to accept. In fact, WSDL 2.0 is beginning to extend itself as a way to provide a service as a REST service. Here's how to define the URL, GET, POST, PUT, DELETE. Here's how it maps back to the service described in WSDL. I think that could be pretty powerful. Now, you can start using your RSS reader as a Web services client. Also WOA is a very compelling idea for building a simple Web platform. So if you think about what's going to be the next generation, I think it's going to be less not more.