News

Enterprises Not Ready for UDDI

• By Matt Migliore
• September 23, 2002

While UDDI was introduced along with Simple Object Access Protocol and Web Services Description Language as a core component of the Web services architecture, the barriers impeding its path into the enterprise have proven to be somewhat higher than for SOAP or WSDL.

Industry insiders say design flaws in the first two versions of UDDI have, in part, hampered its adoption rate. And, some argue enterprises just aren’t ready to ditch their current registry systems in favor of UDDI directories.

Uttam Narsu, a vice president with the XML Web services team at Giga Information Group, says, “UDDI will have a future, but there were a number of deficiencies in the first spec.”

“UDDI 1.0 didn’t even support WSDL very well,” says Narsu, which proved to be a glaring oversight since UDDI relies on WSDL to define the Web services listed in a given registry.

According to Narsu, UDDI and WSDL are now more closely tied, but he says there were other problems with the original version of UDDI, as well.

The most obvious, he says, is that UDDI initially did not allow for the creation of both public and private registries. The flaw, says Narsu, is a relative oddity given that UDDI is based on DNS – a method for establishing computer networks – which has functionality for both public and private environments.

Though many of the early problems with UDDI have been rectified, Narsu believes poor design has set the UDDI specification back considerably in terms of uptake.

He also attributes the technology’s unimpressive adoption rate to an overly optimistic implementation strategy. He says, “There are dozens of registries in a company, and UDDI has the hubris to think all of this can be handled in one place.”

According to Narsu, the notion of a “universal” directory has made enterprises hesitant about UDDI. He says, many companies already have large investments in directory technologies, like LDAP servers, and they don’t want to have to build a whole new server for UDDI.

Earlier this month at the Windows Server DevCon event in Seattle, developers showed just how wary they are of UDDI. At a session called UDDI Services for Enterprise Architects, Rosa Thomas, a program manager for the UDDI group at Microsoft, asked audience members, “How many of you are using UDDI right now?” Only a few people raised their hands.

By most accounts, developer apathy toward UDDI is an industry-wide issue. And many industry analysts have openly wondered whether UDDI will ever become a prominent directory technology. But, at its DevCon event, Microsoft seemed convinced of the power of UDDI, as Thomas urged her audience to “register your services within UDDI” and “encourage developers to discover them.”

Giga’s Narsu also remains optimistic about the future of UDDI. He says, “UDDI is certainly useful in a design context.” But, he believes the timeline for UDDI is a bit longer than originally expected.

Narsu says, in order for UDDI to start making waves in the enterprise, companies must first change the way they think about business relationships. Or, he says, the value of UDDI must be positioned in a different way.

Right now, says Narsu, companies aren’t going to use a UDDI directory to facilitate the dynamic integration of a Web service into their business activities. He says, companies rely on trusted business partners for those sorts of relationships. And, while the technology may be capable of facilitating dynamic integration between Web services, Narsu says the people aren’t ready for it. “The technology barriers for adoption of Web services are being overcome,” he says. “The human barriers aren’t.”

One key move Narsu feels may ultimately help speed the uptake of UDDI, is the recent transition of the spec from Microsoft and IBM to the standards body OASIS (Organization for the Advancement of Structured Information Standards). The move puts the development of UDDI under the control of an independent party, which Narsu says could give the specification more credibility in the eyes of users.

Still, Narsu expects it will be some time before UDDI is prevalent within the enterprise.

“It’ll probably be eight years before UDDI gets deployed on a widespread scale,” says Narsu. “There has to be a level of comfort, and one way to get that is to deploy internally – that is starting to happen.”