Attendees at the Borland conference here had ample opportunity to sharpen their Web services knowledge—and fine-tune their skepticism about adopting and deploying the technology for anything but internal or demonstration projects.
The conference consensus: Business-to-business Web services, especially those involving process integration across more than one companys systems, are not a "this year" proposition for companies that want to move with—not ahead of—key standards, such as those for transaction security.
Even so, development teams should acquire needed skills and should choose and learn new tools. If nothing else, the rapid embrace of Web standards protocols by commercial software vendors is easing the task of EAI (enterprise application integration). This year is "the year of experimentation for companies that dont want to become laggards. In the short run, Web services technologies offer low-cost EAI," said John Meyer, an analyst at Giga Information Group Inc., in Cambridge, Mass., during one panel discussion.
The most unrealistic expectation, Meyer said, is what he calls "the Web services hairball." Developers must learn their lesson, he said, from the past debunking of what he called "the myth of the ORB." Early proponents of Object Request Broker technology implied, Meyer said, that sticking components together was enough to create an application. "That wasnt true, and the equivalent proposition for Web services isnt true, either," he added.
Many enterprise development teams are likely to embark on pilot projects to establish their direction and to assure top management that theres reality to back up the hype. Well and good, said Bill Conroy, e-channel delivery manager of Bank of America Corp.s Global Corporate and Investment Bank Division—but only if those involved can protect themselves against what he called the common mistake of losing sight of the purpose.
"The goal of a pilot project is not to produce a deployable application," Conroy told his technical session audience. "The goal is to capture metrics. A successful pilot project may prove that the technology is worth introducing, even while teaching you how not to use it."
Conroy added, "Introducing the technology is one outcome, fielding the pilot project application is another outcome. But if you cant afford to throw the whole project away, youve made a de facto decision to introduce the technology and deploy your first attempt to use it before you even begin."
Finally, developers were urged to decide for themselves how new tools can solve their real problems, rather than buy into vendors positioning du jour. "Microsoft [Corp.]s .Net is a lot more than Web services," said President and Chief Technology Officer Alain Tadros of Kazoo Software Inc., in Santa Clara, Calif., "but Web services are big now, so thats what you hear. If everything tomorrow is tomatoes, then .Net is all about tomatoes."
Meanwhile, Tadros said, a text-based Web services protocol such as Simple Object Access Protocol "is definitely slower than other distributed systems protocols, so know when to use it. Its not the answer to everything."