Services For Sale

Nearly four years ago, I sat at the back of a packed conference on something new and exciting called Web services.

Web services was going to be bigger than the Web itself. Any machine would be able to talk to any machine, and eventually most apps would be built from components strung together across the Internet. As part of the revolution, why shouldn't enterprise customers become Web services vendors?

But IT had other priorities, like slashing costs. And Web services mainly became a cheap integration method. But recently those giddy early days came rushing back when I spoke with Infravio CEO Jeff Tonkel about his X-registry product, an enterprise registry and repository for publishing and even selling Web services.

Before Tonkel's tenure, Infravio's foray into the Web services market included both development and migration tools. Tonkel then moved the company to the broker space, where Web services is an EAI replacement with performance management and measurement capabilities. But ultimately, BEA, Cisco, Microsoft and the other big infrastructure players are going to own this space. Now Web services/ service-oriented architecture asset management is the center of Tonkel's strategic  vision for Infravio.

As luck would have it, travel giant Sabre needed just such an application. Infravio beat out its competitors because its X-registry is similar to a searchable e-commerce catalog that holds detailed descriptions of services and, more importantly, provides control and approval mechanisms. Sabre decided it was easier to set up shop using X-registry than to build a similar app itself.

Those who know a little about Web services may wonder: Why not just use Universal Description, Discovery and Integration (UDDI)? Mainly because UDDI as it stands is really a spec for a relatively simple  directory and (unique among the basic Web services standards) has lost traction rather than gained it. And the ebXML registry spec, once championed by IBM and Sun, never really got off the ground.

Infravio has no direct competition as yet, but I imagine a few companies may want to enter the space. The great thing about Web services is that it's been a grassroots effort and has lowered the cost of integration.

The problem with it is that developers tend to use it as an ad hoc solution and document it poorly - the key exceptions being the public-facing Web services, such as those offered by Google or Amazon.com. True, what Google and Amazon.com offer is pretty simple, but it's easy to underestimate the effort involved in making Web services reliable, self-service, scalable entities that pretty much anybody can use.

Throw in the proper rights and permissions mechanisms, and that philosophy should also underlie Web services inside the firewall. It's going to take years before the swirl of draft Web services specs settles down, if ever, and even if it does I can't imagine a day when Web services will run around connecting with each other dynamically,  without human intervention. In human-readable form, registries and repositories must capture all the relevant information needed to contract with a Web service, or much of the Herculean effort involved in creating a service-oriented architecture - which expands organizations' integration possibilities by a magnitude - will go to waste. And these registries should include descriptions understandable by business types, not just technologists.

Who knows? Once you've established that sort of repository inside your organization, it's not that big a step to consider selling a few select services over the Internet. At the very least, if you pitch it right, the prospect might score a few points with the business guys. CIO