UDDI seeks its place

Is UDDI (Universal Description, Discovery and Integration) a bust or is it just in an embryonic stage? These questions are being pondered as the Web services directory specification continues to evolve.

Adoption rates have not skyrocketed as some expected, prompting many to re-examine where UDDI will fit into the Web services puzzle.

The UDDI standard is intended to provide central directories where consumers of Web services can access various services, either within a company’s firewall, via an extranet, or on the public Internet. Service providers can register them and make them available via UDDI, which is based on technologies such as XML, HTTP, and DNS.

Companies can set up these registries internally and choose to extend access to partners. Microsoft Corp., IBM Corp., Sun Microsystems Inc., and Systinet are among companies that offer, or plan to offer, UDDI products. There are also public UDDI registries deployed by Hewlett-Packard Co., SAP AG, and Microsoft.

But experts say that adoption of UDDI has been slow-going, with users mired an in an educational mode and many still ramping up their UDDI and Web services efforts. The concept of UDDI is still maturing, according to Joel Munter, a senior software engineer at Intel and the company’s lead representative to the industry’s UDDI effort.

“Things are ramping, and they’re ramping as the industry drives it and as the industry matures in its run-time use models,” Munter said in an interview late last month.

Munter added that tens of thousands of services are accessible in public registries. However, loftier expectations in some circles had been for hundreds of thousands to millions of accessible services.

UDDI enables services such as allowing a frequent flyer of an airline to access that airline’s support services in a registry via devices such as a laptop, PDA, or cell phone, Munter said.

“In mid-2003, you’re going to see an awful lot more Web services in use,” but the next 12 to 18 months will continue to serve as an adoption phase, Munter said. UDDI adoption will likely pick up as Web services deployment does.

Version 1 of the UDDI specification was announced in September 2000 and Version 2 followed in June 2001. Associates of uddi.org, including Munter, are preparing to release Version 3 in July, focusing on features such as communication between private, semiprivate, and public registries. Security issues also will be addressed.

Members of uddi.org also are preparing to turn over jurisdiction of UDDI to a yet-unnamed standards body this summer. A request for proposal document has been published and sent to a selective number of standards development organizations, Munter said.

During a panel session featuring uddi.org members at the Software Development Conference in San Jose, Calif., in April, panel moderator Brent Sleeper, a consultant at The Stencil Group, charged that registries were “missing in action” and that few services were available.

But Microsoft’s public registry has “thousands” of services registered in it, according to Jenna Miller, a business development manager at Microsoft who has been part of the company’s UDDI team.

Miller acknowledged that enterprises are currently focusing on internal deployment of Web services before extending access externally.

“Customers see huge opportunities for using the public registry, but the first thing is to get their house in order,” Miller said.

Microsoft will release its Enterprise UDDI Services software, for deploying private UDDI registries, as part of the Windows .Net Server operating system due within a year, according to Miller.

IBM’s Bob Sutor, director of e-business standards strategy for the Somers, N.Y.-based company, referred to UDDI as a “chicken and egg” situation: Its growth depends on development of Web services. Companies are deploying UDDI within firewalls first; extending beyond that will occur as enterprises start using UDDI for integration, Sutor concurred.

IBM included support of private UDDI registries in its recently announced WebSphere 5.0 application server.

Sun’s upcoming UDDI offering will be based on its Sun ONE (Open Net Environment) Directory Server, and will enable setting up of private registries for publishing Web services, company officials said.

Novell also planned to update and resubmit in late May a specification entitled “LDAP schema for UDDI,” originally submitted to the Internet Engineering Task Force (IETF) in February. The spec defines a standard format for representing UDDI data types in an LDAPv3 directory.

The company is looking to highlight the similarities between UDDI registries and directories as repositories for Web services, as well as the authentication and security features of directories that could supplement UDDI registry security.

Systinet has offered a UDDI server called WASP (Web Application and Services Platform) UDDI since October. The product is available free for download and testing. The company has seen the interest level in WASP UDDI fluctuate over time.

“We had very little interest until about January, then we had a remarkable uptake in downloads,” said Anne Thomas Manes, CTO of Cambridge, Mass.-based Systinet.

The spike occurred, Manes said, because Web services have begun to reach “critical mass,” which may bode well for future adoption of UDDI.

Ed Scannell, InfoWorld editor-at-large, contributed to this report.

Have Patience, UDDI is Coming

We in the Test Center have been beating the drum for Web services for some time now. But Rome wasn’t built in a day, and I’m not wondering why “it,” as in the mass adoption of Web services, hasn’t happened yet. That’s because I see it happening around me, as an evolutionary, rather than a revolutionary, process.

The graphic Web browser was revolutionary, but what we’re going through now is simply the natural development of working, usable Web services implementations.

Take UDDI (Universal Description, Discovery, and Integration), for example: the first swing at a standard came less than two years ago, and a third revision is due later this summer. In the meantime, major players including IBM, Microsoft, and Sun have forged ahead on their own, for better or worse.

I just don’t have much patience for whiny panellists who complain that users and vendors aren’t moving fast enough on UDDI.

Hey, guess what? Recessions (or whatever you want to call this generally crappy economy) aren’t the time to invest in technology for technology’s sake. Businesses are having enough trouble just keeping the lights on with skeleton crews. They certainly aren’t looking to complicate their IT puzzle any further.

The vendors are pushing ahead, but to no one’s surprise, they also have their own agendas to advance. I understand why Microsoft wants to tie its industrial-strength UDDI server into Windows .Net Server; after all, Active Directory has to be good for something besides Exchange.

But I would be surprised if it makes the final cut for Windows .Net Server, since I expect to spend the summer months playing with Beta 3, and new features become more ill-advised with every passing day. If they aren’t in that beta release, UDDI services will most likely show up as a downloadable option, and later be included as part of a service pack.

In the long run, UDDI’s evolution may well absorb LDAP into a new standard for directory services. At least, that appears to be the aim of Sun’s efforts. Whether or not anyone nibbles at Sun ONE (Open Net Environment) remains to be seen, but it’s clear that the lines are already being drawn. Ultimately, customers will vote with their dollars.

I’m also a little put off by the complaints that public services aren’t available. Of course businesses are starting with internal UDDI registries -because if your own house isn’t in order, how can you expect your partners’ to be?

The idea that there will be one master registry is foolish. “Federation” is the magic word here: UDDI will succeed only through voluntary association of businesses exposing their registries for public consumption.

Businesses aren’t ignoring UDDI, they are merely figuring out how best to take advantage of it. Even if the number of available services is a tenth of what we thought it would be back in 2000, that explosive growth will still happen. It may simply take a little longer, that’s all.

Anyone complaining because a technology takes more than two years to go from an initial standard to widespread use needs to learn the virtues of patience.

Personally, I’d rather see businesses take their time, do UDDI right the first time and move on, than see a repeat of the ERP (enterprise resource planning) and CRM disasters that too often absorb an IT department’s entire attention.

– P.J. Connolly