Sun Microsystems' JavaSoft division has been successful in promoting Java as a way of delivering client applications or applets over the Internet, and as a programming language. Now the company is pushing it as a means of deploying server-side component software. Niall McKay and Ted Smalley Bowen talked with JavaSoft president Alan Baratz about how the company intends to make a buck from Java's successIDG: A hot topic in 1997 was Web application servers with Netscape's acquisition of Kiva and Novell's licensing agreement with WebLogic. Will you license or buy a Web application server or Java application server technology?
Baratz: I don't think so. Our approach is a little different. First, the operating system is a Java Virtual Machine.
Then we have synchronous local, which we call JavaBeans, and synchronous remote, which we call Remote Method Invocation. We're not building our own asynchronous inner-process communication mechanism, but we have the Java Messaging Services.
We're not building our own database management system, but we did [build] Java Database Connectivity APIs. We're not building our own directory servers, but we did [build] the Java Naming and Directory Interface APIs. We're not building our own [transaction processing] monitor, but we have the Java Transaction Service.
So, we're building up an alternate platform that consists of the Java virtual machine and all the class libraries to support the Java intercommunications protocols, and the mapping to whatever middleware services you happen to have installed. I would call this platform an application server.
So what other products or services will JavaSoft develop and market?
That's to be determined. I will tell you that there is high probability that in the very near future we will add calendaring because currently there's no standard calendaring service. Mail has standard IMAP4 and POP3, so you don't really need to build your own mail servers. But one of the questions on our mind is, what happens in the small and midsize businesses that can't afford to go in and buy a CICS or a Tuxedo?
What do we do? We are having internal discussions [about whether we should] put a low-end TP monitor into Enterprise JavaBeans. It would always be low-end, and there would be a simple migration path.
Is there a similar opportunity in the messaging server space?
Potentially for messaging. Potentially, but probably not for database.
There are various rumours that JavaSoft is now under a lot of pressure to produce revenue, is this true?
JavaSoft actually did quite well last fiscal year. We came in at 25 per cent more than our target. We're doing well this year. I think that our goal is to build a viable organisation within Sun Microsystems.
A lot of times people ask me: "Are you about ubiquity or about revenue?" But we're actually about ubiquity without leaving money on the table.
Our goal is to make the Java platform rich, stable, and ubiquitously available. But there's a value that's been associated with it, and there's no reason not to leverage that fact and extract some of that value.
So how are you going to recoup your investment costs?
We now have more than 100 licensees of the Java platform. We are now moving into an end-user binary product.
Have you cordoned off or identified areas that might compete with your partners?
There may be some competition with some of our ISVs in mail and calendaring, but I don't think that will be a big concern. We have agreed that we're going to work closely with IBM, Lotus, and Oracle on the WebTop.
The thing you need to keep in mind, though, is that unlike OS/2 or Windows or most other platforms, there isn't one company that really has all the intellectual property for the Java platform.
We may choose to move up in the application space, but we don't have any particular advantage in that space over any other large company. That having been said, we are not going to push into a broad range of applications because we do understand that it is critically important to build a base of ISVs that are successful in Java.
So you don't see Sun competing with these ISVs?
Quite the contrary, we're spending a lot right now, trying to help third-party ISVs to be successful. So we're trying to very carefully select areas that don't step on the toes of ISVs. Sometimes we make a mistake. I mean there are a couple of ISVs that have done base infrastructure stuff, such as Novera.
That's hard for us because it is doing things that we know need to be a part of the platform and that all our licensees and developers are telling us need to be a part of the platform.
I mean, we've got a moral and practical commitment to working with other partners out there.