Keller: No, I don’t think it’s been overhyped. I think the challenge is it’s still early in Web services and building the interfaces for them. I think the frustration you’re hearing is we’re still learning how to build these interfaces, how to build them well, how to build the tooling for them, how to make them practical for applications. And it is why we build a set of technologies and deliver them in something called the Java Web Services Developer Pack, which [features] APIs and our Web-services support. Any developer can take and download [that for] use in their IDE and their application server and their Web server in their Java SE [Standard Edition] runtime, in their Java EE [Enterprise Edition] runtime to allow them to get access to the latest and greatest implementation of the Java APIs for doing Web services. One of the things we just put into the Developer Pack is an acceleration technology. One of the frustrations is people would like Web services transports to go faster. This version of the Developer Pack we shipped this week has a binary encoding capability that is based on standards from the ITU that allow us to encode a binary form of the communication and send that between nodes at a faster pace. Our benchmarks to date show anywhere from a 150 to 400 percent [increase]. And so that helps us address some of the frustrations that people see.
IW: Why should developers build Web services on Java rather than another platform, say, .NET?
Keller: I think people should build things in Java -- not just Web services, but all application types -- for its portability, its productivity and its, if you will, scalability of application. [Developers are] able to go from the smallest devices, like Java Cards and phones and DVDs, and have that same code be available and running on the larger devices, whether they are desktop computers or servers. It really does give you the kind of return on your investment as a developer. You’re investing your skills and I think the most productive and highest return is being able to apply those skills over the broadest range of deployments. I think that’s why people choose Java.
IW: Sun has open sourced a version of its application server as well as an enterprise service bus and Solaris. What else might be open sourced at Sun?
Keller: At this point those are the ones we’ve been able to get to the point where we know open source will help those communities. We’re still in the process of evaluating the rest of the Java Enterprise System stack and what might be next. We haven’t really decided yet. But what we look for is communities that can benefit from having source-code access. What we’re trying to enable is the ability of the developer to understand how to write better applications with that technology, of being able to base their research and innovations on our technology, of being able to have control over their deployments so that if they need to fix something based on time requirements, they have control. And ultimately there’s a group of developers we believe who want to actually contribute to the development of the underlying platforms. So that’s what we’ll use as a set of requirements to evaluate against each of the projects and when we find the right kind of technology that has that kind of requirements, that’s when we move them into open source.
IW: Would you say Sun is still primarily a hardware company, generating far more revenue selling servers and workstations than selling software?