FRAMINGHAM (07/07/2000) - IBM Corp. wants to transform its WebSphere application server into an infrastructure platform and double its share of the Java application market. But analysts and users said this week that IBM still has some work ahead of it.
As part of its newly unveiled strategy, IBM will embark on a yearlong, $1 billion development and marketing project to convert WebSphere into an Enterprise JavaBean (EJB) middleware platform from which customers can more easily run prebuilt Java applications.
At PC Expo late last month, IBM released WebSphere Application Server 3.5 and announced two applications that run on top of it: WebSphere Portal Server and WebSphere Personalization Server.
Nick Gall, an analyst at Meta Group Inc. in Stamford, Connecticut, said bringing applications together under one WebSphere framework makes sense.
"Yesterday, WebSphere was just a JavaBean server with raw programming power; you could build a lot with it, but it had no prebuilt functionality for portals, tracking or shopping carts," Gall said. "Now, they're doing hardware engineering to rewrite [WebSphere] to run on top of a Java platform, but they still have their work cut out for them."
Tom Guyette, a senior systems analyst at the Federal Reserve Bank of Minneapolis, said IBM's WebSphere currently requires a lot of custom coding to add new capabilities, but he would avoid moving to the new WebSphere if the migration process were too complex.
Now, "you have to talk to a database with an EJB, and you have to create that interaction in VisualAge [IBM's application development tool] and then transport it to WebSphere," said Guyette. "But I want to minimize any changeover. If they come up with bundled solutions, it would help, but we still have to have more detail."
According to Cambridge, Massachusetts-based Giga Information Group Inc., IBM trails San Jose-based BEA Systems Inc. in the EJB application server market.
BEA last year held 32% of the market, compared with IBM's 16%. But Giga predicted that IBM's share will grow to 24% by year's end, while BEA will drop to 24%.
"BEA had a tremendous time-to-market advantage and had viable product in channel by April 1999," said Mike Gilpin, an analyst at Giga. "It was much later in the year before IBM had an effective product. But when you look at buying behavior, IBM seems to win as often as BEA, and sometimes more often due to [price breaks]."