Are Java magic bullets on target?

Technologies that solve Java problems work - mostly.

By Chris Nerney
Network World, 7/6/98

Ever since it was introduced three years ago by Sun Microsystems, Inc., the Java programming language has been burdened with outsized expectations and major questions regarding its ultimate value in the enterprise.

Specifically, Java critics - and even supporters - cite three problems that have to be overcome for the language to enjoy widespread acceptance in large networks: slow performance, lack of consistent portability and inability to scale.

In response, Sun has unveiled a trio of fixes, or "magic bullets," specifically designed to address each of these issues. What follows is a look at how close each of these magic bullets has come to hitting its target.

Whither HotSpot?
In April 1997, Sun announced with great fanfare its recently acquired compiler technology. Company officials boasted that the technology would double the speed of Java applications, which were chugging along well behind applications written in C++.

That technology, HotSpot, is designed to find the parts of an application that can be accelerated and make them run faster. JavaSoft purchased HotSpot technology early last year from start-up LongView Technologies LLC of Palo Alto, Calif.

Today, however, like characters in a Samuel Beckett play, Java programmers are suffering an angst-ridden wait for this performance savior. Originally due out last summer, HotSpot's debut has been repeatedly delayed, and the commercial release is now scheduled for sometime this fall.

The delay has led to speculation that the highly anticipated technology is having problems.

And that could spell trouble for Java. Even though third-party vendors such as Tower Technology Corp. of Austin, Texas, have devised their own speed solutions, some potential customers are waiting for Sun to find a solution for Java's performance woes before embracing the programming language.

"There is nothing more critical for Sun to get out right now than HotSpot," said Scott Dietzen, marketing vice president for WebLogic, Inc., a San Francisco-based Java application server vendor.

A plug for portability
From the start, the great attraction of Java has been its "write once, run anywhere" promise. And the biggest threat to that cross-platform notion has been Microsoft's aggressive bid to split the Java platform, essentially creating a second version of Java for Windows. Then last December, at Internet World in New York, Sun unveiled plans to release a product designed to overcome Java incompatibility woes in any browser.

The product, eventually called Java Plug-in, would check a user's machine to see if the browser had the latest Java Virtual Machine (JVM) needed to run Java applications. If it didn't, Plug-in would let users download the JVM to reside permanently on the client.

Viewed by many as a brilliant countermove to neutralize Microsoft's effort to limit Java, Plug-in also is designed to allow users with older versions of Netscape Communications Corp.'s browsers to run Java applications.

The software has been embraced by a number of vendors as a way for customers to use their Java-based products regardless of the browsers residing on their desktops.

Plug-in has security vulnerabilities that make it impractical for downloading Java-based Internet applications, said a systems integrator working with a large telecommunications company that wants to use Java.

"The infrastructure for trusted applets is not there," said David Boydston, president of Solutions Consulting International, a Santa Barbara, Calif., company that also sells a serial port control software product for Java applications. "You have to distribute security modules, and people have to copy them to their disks."

Enterprise JavaBeans (EJB) was designed with one goal in mind: to allow Java to play with the big boys. Introduced late last year - and released as a final specification in late March - EJB was touted by Sun as the tool that would drive Java development to the server in 1998 and shatter the myth that Java was a client-side toy.

EJB features chunks of precoded software that can be joined to build a larger application that can run on a JVM. The component software has been eagerly anticipated by programmers who want to build large, server-based applications but lack adequate tools.

"We're looking to move toward an EJB-based component model on the server," said Deva Hazarika, founder of Moai Technologies, Inc., a Web-based business-to-business auction software vendor located in San Francisco. EJB has attracted support from other major vendors, particularly IBM, which in March announced plans to add EJB functionality to its middleware products. WebLogic this month released a beta version of its Tengah application server that supports EJB.

Still, while the building blocks for creating powerful server-side Java applications are just about here, the first wave of EJB-based applications likely is months away. Until the applications begin hitting enterprise shores, EJB's legacy will remain unclear.