InformationWeek

October 19, 1998

Java's Future

continued...page 3 of 3

Illustration by Eric Mueller

Visual J++ 6.0 is Microsoft's newest development tool for ActiveX controls, Windows executables, Java scripts, HTML, and Java applets on Microsoft platforms. It not only supports Microsoft-flavored Java, but also provides direct access to the Windows platform.

But Bill Dunlap, Microsoft's Visual J++ 6.0 product manager, runs hot and cold on Java. He expects Visual J++ to be used as extensively as C++ and VB as a Windows development tool by 2000, and he'd be happy if 25% of COM development was with Java and 50% with VB. But he puts his emphasis on Java with Microsoft overlays and strategies--COM for distributed computing and Dynamic HTML on the client. He also stresses that Microsoft research shows 3 million developers use VB compared with about 150,000 using Java.

Dunlap dismisses the notion of Java's portability, arguing that businesses buy hardware for a specific reason: "Java is a great language for accessing the great capabilities of a specific machine. It's a pipe dream to think people will move large-scale apps from machine to machine," he says. He argues that Java isn't the development cornerstone of the enterprise, but that its primary use will be as midtier software between the client and legacy app on the back end.

"For performance-critical systems, you want to use Visual C++," Dunlap says. "For the fastest bang-'em-out apps, the environment is Visual Basic, and the way to build Windows and Web apps with a single framework is Visual J++." Dunlap doubts Java performance will ever match that of C++, and says Microsoft is researching how to create a native Java compiler for Windows to improve Java's performance.

What would boost Java in the market? Third-party applications wouldn't hurt. If the enterprise has yet to standardize on Java development tools, it comes as no surprise that enterprise Java applications remain works in progress. Last summer, Sun and IBM pegged the number of third-party Java applications at about 1,200. That's probably one reason that about half the businesses responding to an informal Network Computing magazine E-mail survey give Java a grade of D or F today when it comes to availability of enterprise applications. About 8% say Java will continue to be below average by 2000.

Lew Tucker, director of strategic relations for Sun, says Java development is much further along than the 1,200-app statistic suggests because that figure doesn't reflect the use of Java as an integration platform for packaged enterprise applications such as those from SAP and PeopleSoft, or the use of Java to build customized applications.

But the issue isn't that enterprise applications won't ever be available--it's that they can't come soon enough. John Rymer, president of Upstream Consulting, says he expects enterprise-ready products to begin arriving by early 1999.

Another factor that will determine Java's long-term success in the enterprise is the establishment of a single reference model for Enterprise JavaBeans. Without such a model, users say, there's no guarantee of EJB interoperability.

Users and analysts say an EJB reference model has yet to emerge simply because Sun is overwhelmed by the breadth of its Java undertaking, coupled with the company's recent reorganization. Analysts speculate that because Sun's internal product development is complemented by the control it exerts over a reference implementation, it will resist farming out so vital a task. That, in turn, will slow Java's acceptance in the enterprise.

So what's Java's future? Because Java is only three years old, Java veterans and proficiency won't pop up overnight, even if, as Gartner Group's Smith concludes, "Java is rapidly becoming the teaching language at universities."

Businesses with the largest Java programming teams have attacked the staffing problem by retraining C and C++ programmers and establishing mentorships for programmers with less experience.

But it isn't the cultivation of Java programmers, the creation of a reference model, or even technology investment that will push Java out across the enterprise. Gartner Group's Smith says it's something much simpler: changing the perception among old-line IT staffers that Java is just for the Web.

Christy Hudgins-Bonafield is business and trends editor at Network Computing, a sister publication of InformationWeek.

return to page 1, 2

Illustration by Eric Mueller