Report from the Web services front
Summary
Topics
The event had the feel of the early days of PCs, where industry players discuss an exciting new toy in an almost academic way, wondering what the long-term implications will really be. There was general agreement that Web services lack technical sizzle--they merely make XML integration real through a simple set of messaging protocols. Here are the main observations and issues that floated to the top:
Keep it simple. IT should get rolling with Web services projects now rather than waiting for new security and business-process protocols to emerge. The Simple Object Access Protocol (SOAP) and Web Services Description Language (WSDL) are all IT needs to get rolling with enterprise XML integration projects inside the firewall--which provide the major, near-term opportunity for Web services. If we wait for standards committees to hammer out Web services protocols that duplicate the functionality of existing security and business-process controls, the complexity of implementation could rise to the point where Web services will fizzle like other complicated component architectures (such as the Open Software Foundation's DCE or the Object Management Group's CORBA).
Lower the cost of intra-enterprise integration. Today, many companies rely on pricey enterprise application integration (EAI) products with adapters that enable specific, commercial enterprise applications to communicate with each other. Web services will eventually make EAI obsolete, fostering simple, message-based data exchange among apps. In pilot programs, several vendors claim to have already dramatically reduced integration project costs and time-to-market. Again, simple is best: Just enabling Microsoft and Java objects to talk to each other--an issue EAI doesn't address at all--offers immediate benefit.
Solve real B2B problems. The example most often cited: replace electronic data interchange (EDI). This system of dedicated, proprietary links set up by corporations, typically for ordering huge quantities of raw materials, is too expensive for most businesses. Web services have a chance to succeed where previous B2B initiatives have failed: low-cost machine-to-machine transactions between enterprises. HTTP plus SSL may not be fast or reliable enough for many B2B applications, so creative new solutions, notably from Kenamea and Slam Dunk Networks, provide secure HTTP alternatives.
Don't be distracted by B2B-style hype. Several participants seemed embarrassed that Universal Discovery Description and Integration (UDDI) is one of the three protocols that define Web services. The idea of a master UDDI directory, where thousands of Web services would automatically integrate with other Web services over the Internet, has problems--the most obvious being that capturing reliable quality-of-service information may never be practical. Companies won't switch dynamically among Web services of uncertain origin any more that they formed ad-hoc business relationships in B2B marketplaces. And if you need to set up a trusted relationship first, why bother with a UDDI directory in the first place? The best answer to that question: UDDI directories could be used in private exchanges where the players already know each other, or within a single enterprise to make application integration easier.
A word of caution for those considering a Web services pilot program: Partisanship in the Microsoft .Net and J2EE camps continues to escalate as both try to convince the market, either directly or by implication, that they have the only truly viable platform for Web services. My advice: Develop and deploy with the least disruption to your existing infrastructure. If you have some C++ developers handy and you've already deployed Windows in the back office, snag a beta of Visual Studio .Net and have them build a useful Web services application written in C#. If your J2EE developers are spoiling to wrap EJBs in SOAP, upgrade to the latest, Web-services-enabled version of your current J2EE application server and let them have at it. Experiment using what you've got. After all, one of Web services' chief promises is low-cost development and integration, a welcome idea in tough times.
Are you excited by the integration potential of Web services? Or do you have too much else on your plate? E-mail Eric or Talk Back below.
Talkback - Tell Us What You Think
The best of ZDNet, delivered
ZDNet Newsletters
Get the best of ZDNet delivered straight to your inbox
