Last week, at a roundtable sponsored by the elite PR agency Blanc & Otus, I realized that we're at a similar stage with Web services. Many millions in marketing dollars spent by Microsoft, IBM, and their competitors have raised expectations for an infant technology to an absurd degree, when so far we have little more than a handful of protocols and general agreement that easy machine-to-machine communications via XML would be a very good thing.
Execs from Sun, TIBCO, EDS, E.piphany, Grand Central, Alphablox, and a few Web services startups eyed each other around the table. The evening started with a bang, as Sun ONE VP Marge Breya suggested that Web services could be "the fundamental change of this century" because--just as mass production transformed manufacturing--Web services will provide an "assembly line for business processes." This is actually a brilliant distillation of the idea that business logic can be completely abstracted from the platform it runs on, a future vision where each business process can be a reusable, interchangeable part that will run on any system, anytime, anywhere.
But as the evening wore on, the journalists around the table cranked up their spotlights and asked for specifics. Tell us what Web services are really good for, they demanded. Because that question is similar to asking, "What can you do with HTML over HTTP?" the participants naturally tended to give hypothetical examples relating to their own business. Philip Fernandez, executive VP of products for E.piphany, speculated about the effect of Web services on CRM. Alphablox CTO Pete Hirsch offered soaring predictions about his company's specialization, business intelligence. So it went.
And when will Web services take off? Well, one thing was clear: They haven't left the hangar yet. Around the table, even pilot programs that met the basic qualification for Web services (XML and SOAP over HTTP) were somewhere between rare and nonexistent. To be fair, the roundtable lacked representatives from Microsoft and IBM, either of which could have ponied up some Web services case studies (though everyone knows those deployments were implemented expressly to prove that Web services work).
After the show, Alphablox's Hirsch gave me a succinct summation of where we're at with Web services: They need to be supported by platforms, tools, and customers, and so far, only the platform (an enterprise-class Internet infrastructure with SOAP-enabled server software) is really in place. Tools from major vendors--such as Sun's Java Web Services Developer Pack or Microsoft's Visual Studio .Net--are just arriving. And customers? We're really talking about 2003.
To move beyond limited implementations, those customers will face significant hurdles. Creating a truly modular enterprise application infrastructure is a huge undertaking that will demand much more sophisticated application management. The verbosity of Web services protocols will require major hardware upgrades. Even the basic Web services protocols may need some more work. Moreover, as Hirsch noted, it's important to remember that Web services protocols are indeed just protocols, not APIs; to enable two applications to swap data, you need to agree on (and perhaps develop your own) specific XML schema. And for real integration outside the firewall, the industry will need to standardize on common security, authentication, and business process protocols. That process has just begun.
My point is not to say that the Web services trend has flagged or that the potential is anything less enormous than Sun's Breya says it is. Eventually, Web services will provide spectacular integration benefits and new Internet-based applications that boggle the mind. At this point, who knows what form they may take? Yet the road to big payoffs will be long and hard--and most industry players, who know full well the agony enterprises went through implementing ERP systems in the '90s, don't want to dwell on the up-front costs.
Yes, the tools are here and it's time to start playing with them. You can use them now where Web services make obvious sense, such as quick integration of J2EE and Microsoft applications. And as you look ahead, by all means get serious about XML coupled with component-based app development and management.
Meanwhile, be wary of vendors or service providers who, under the glare of unrealistic expectations, promise overnight miracles or fudge the definition of Web services for their own purposes.
Are you ahead of the Web services curve or behind it? Prove me wrong and tell me all about your successful Web services pilot programs. E-mail Eric or Talk Back below.
