With remarkable frequency, people who tout Web services as The Next Big Thing always begin by trashing wireless. Wireless requires massive investment in infrastructure; Web services use current infrastructure. Wireless offers expensive solutions to problems that never existed; Web services offer cost-effective solutions to everyday IT problems.
The message is that the Web services concept stands apart in its common sense. It’s a simple idea: Enterprise applications should be broken down into reusable components called services, each one performing a distinct task. These services can then link together across or within enterprises using the only data exchange standard everyone can agree on, XML. Like a kid stacking toy blocks, you can build applications from Web services very quickly, borrowing blocks from others when you need them.
Drawing on systems as disparate as a Hitachi mainframe and an NT server, for example, insurance company execs could click and drag Web services to create one-off packages. Car rental and airline reservation systems could integrate across the Internet with minimal development effort. Once you break applications into Web services, low IT overhead makes such experimentation practical?or so the theory goes.
“We’ve been talking about modular and object-oriented software for years, and in many ways this is nothing more than the realization of that,” says Frank Moss, chairman and cofounder of Bowstreet, an enterprise portal provider that is building its business on Web services. “It’s finally happening due to the emergence of XML as the standard for interoperability. All computing is going to become Web services.”
A statement like that would stretch credibility but for the all-star cast supporting it, including HP, IBM, Microsoft, Oracle and Sun. Predictably, each vendor has a self-interested spin on Web services. And the Web services world has already split in two, with Microsoft’s .Net promoting services that run only on Windows while the J2EE (Java 2 Enterprise Edition) crowd pushes its own platform-agnostic, Java-specific solution.
Yet the chorus of heartfelt endorsements can’t be ignored. IBM has even deployed its own Web services evangelist, Steve Holbrook, who cheerfully asserts to anyone who will listen that Web services “is going to happen more rapidly than we’ve seen integration efforts happen before.”
Weaving the Services Web
Holbrook’s optimism stems in part from the accord on key Web services standards by industry heavyweights who normally agree on very little. Such harmony is essential, because XML by itself merely describes a broad framework. Three, more narrow standards?Simple Object Access Protocol (SOAP); Web Services Description Language (WSDL); and Universal Description, Discovery and Integration (UDDI)?enable Web services to talk to one another. (See “Web Services Standards,” this page.)
Created by Microsoft, IBM and others, both the SOAP and WSDL standards are in their 1.1 versions and have been submitted to the W3C standards group. Although in its second version, the UDDI spec won’t reach standards organizations until mid-2002. And a host of other proposed specs for secure transactions, digital signatures and encryption could be bouncing around committees for some time.
Service Without the Smile
When will Web services take off and what, if anything, should you do about them now? Here the happy face of industry agreement cracks. The answer you get depends on which vested interest you ask?and those companies that are further along, such as Microsoft and IBM, will tell you to take action sooner.
“I think Microsoft is way ahead in Web services?over Sun, in particular,” says Gartner Analyst David Smith. “They have seen the future, and it is Web services?and they have pointed the company in that direction. Sun has not done that. Sun was in denial, and they are starting to realize they’d better do something or lose the leadership that they have had.”
However, even Microsoft acknowledges that enterprises won’t start building .Net Web services en masse until next year or the year after. Meanwhile, Microsoft’s Charles Fitzgerald, director for .Net platform strategies, suggests that enterprises “pick a pilot project and start to work with XML Web services.” He cites Dollar Rent-a-Car’s SOAP-based integration with Southwest Airlines as an early, shining example.
Sun Microsystems Chief Technologist of iPlanet Products Hal Jespersen takes a more conservative view. “I think that the infrastructure that’s available today is not suitable for real-world business applications yet,” he says. “Some of the standards are in the category of emerging. I’ll give you an example: SAML [Secure Assertion Markup Language], the security standard that passes security credentials from one enterprise to another. That’s in the standards committee, and I would expect that to be done by the end of the year, probably.”
Even the tools to create Web services stir controversy, perhaps because it’s here that Microsoft’s lead seems most obvious. The Visual Studio.Net development environment, due to ship by year-end, includes XML- and SOAP-enabled versions of Visual Basic, C++ and C#?C Sharp, Microsoft’s new Java-like programming language for .Net development. The Java community likes to slam the forthcoming suite as incurring a steep learning curve for current Visual Basic programmers, and has nothing but scorn for Microsoft’s Java killer, C#.
Scott Dietzen, CTO of BEA Systems’ E-Commerce Server Division, makes a stronger point when he compares the amount of effort that will be required to convert Enterprise Java Beans to Web services versus rewriting Microsoft distributed components: “The expectation is that a 30 percent to 60 percent rewrite will be required if you’re going to take an existing Visual Basic application and turn it into a .Net application or take an existing COM+ component and turn it into a Web service.” By contrast, he says, “Every Bean application that’s developed in Java is already enabled for Web services in the model that BEA is supporting.” For now, that model means you must run the latest SOAP-enabled version of BEA’s WebLogic application server.
As Sun’s Jespersen notes, the mechanics of building Web services matter far less than sound architectural planning. Asked what IT execs should do now, Jespersen offers this insight: “I would have the architects in my organization take a look at what applications I have today and see how to decompose those applications into Web services,” he says. “I believe the heavy lifting associated with designing a good Web services infrastructure is to do that analysis?it’s not to wrap SOAP around it.”
Some lifting, heavy or otherwise, will be required to transform even the most component-oriented IT shop to a Web services architecture. Architects will have to think hard. Developers will need to be reeducated. And many IT managers will balk at the idea of relying on distributed components of uncertain quality over Internet connections that although increasingly dependable, can’t match the availability, performance or security of in-house systems.
Again, the unanimity among industry giants in answering such objections is striking. Almost everyone sees Web services as inevitable; the disagreement is in the timing. Few argue with the strategic advantage: the ability to rearrange Web services dynamically into custom applications that map closely to business processes or customer needs?without breaking IT’s back. Although the big players so far have focused on Web services development, several smaller companies?including Asera, Avinon and Bowstreet?are producing graphical tools that the business side can use independent of IT to build working applications from Web services components.
Like Bowstreet, Asera was delivering Web-services-like solutions years before the phrase was hot. Asera’s eBusiness Operating System lets developers create Java-based “logical building blocks” using functionality offered by monolithic software from Ariba, i2, PeopleSoft, SAP, Siebel and others. Nonprogrammers then employ the Asera Development Workbench to reshuffle those services to their heart’s content. Avinon’s solution, the NetScenario Studio graphical environment and the NetScenario Business Server, today provides an arena for manipulating .Net Web services?and if everything goes as planned, J2EE Web services by early next year.
Where will Web services see their first practical applications? Microsoft’s Fitzgerald suggests that business processes with the most points of integration with other systems, such as CRM, will benefit most. Gartner’s Smith believes the killer Web services app for B2B will be the private exchange, where trusted trading partners integrate via XML without relying on the still evolving UDDI directory. And Bowstreet’s Moss sees the ability to mass customize portals for customers and internal operations as the main market driver.
Each enterprise will develop its own priorities based on its own pain points and opportunities. “Companies are transitioning from being product providers to becoming service providers,” Moss says. “To grow their revenue, enterprises have to provide more and more services around their products or existing services. The Web services architecture is a natural way to do that. If you were to try and provide those services using programmers to build applications every time you wanted to bring a new service to market, you’d never make it.” n