The development of XML promised a great deal: the unification of all enterprise software applications that can talk to each other, all seamlessly linked on new enterprise platforms.

This new integrated transaction and software infrastructure has been dubbed 'web services'.

The promise is that these web services will be easy to access on demand via the internet, requiring no special arrangement between trusted partners which can build new and innovative ways of exchanging data and carrying out transactions.

At the same time, web services could make obsolete many traditional enterprise application integration (EAI) services by forcing the use of reusable, well-documented interfaces in IT systems.

Some basic web services are already in everyday use. Major UK supermarkets such as Tesco and Sainsbury already have web service-linked loyalty card schemes, where basic web services are used to process requests either online, between retailers, through the checkout or via in-store terminals.

There are plenty of theories about what web services will do tomorrow, but we hear little on what it can achieve now; the original hurdle of defining and accepting XML has not been cleared.

With XML, there was a need for a transport mechanism, which became known as Simple Object Access Protocol (Soap). Then Soap needed a standard form of message to carry, hence Web Services Description Language (WSDL).

So that users could locate WSDL services, the Universal Discovery Description And Integration (UDDI) mechanism was created.

These steps are not unlike the evolution of the web, starting with the definition of HTML and the standardisation of the HTTP communication protocol.

But web services does not yet have its killer app. XML was over-hyped, as if it were a new alphabet promising that great works of literature would immediately follow once the letters were accepted as standard. Yet so far, most of the early adopters are keeping their projects small-scale and internal.

"You see a lot of work going on behind the firewall," said AMR Research director Eric Austvold.

That's a common refrain from even those in the web services development and infrastructure business, although they are quick to advertise the benefits of implementing web services integration internally.

"What it has already started to enable is better business automation inside a company across departmental boundaries," explained IBM programme director Steve Holbrook.

"People usually have concerns about security, and the standards associated with security and web services are not complete yet."

Integration of web services
One of the major pushes for web services has been the promise of easy application integration by creating Soap/WSDL-based front ends to any number of legacy data sources and applications, which can then pass messages back and forth in a web services environment.

The idea upset the EAI traditionalists, and even those in favour of web services admit that the technology is not a drop-in replacement for more conventional integration.

"There are established, defined limitations," said Gartner research director Whit Andrews. "Nobody should look at Soap and say that from now on they are going to use it for all application integration projects.

"It is most useful for look-and-listen integration, the opportunity for one application to look at another's data or to monitor activity and behaviour."

Austvold added: "It puts pressure on EAI vendors. They become niche players, and you're going to see more companies leaning to buy from a single provider than from these independent players today."

Major technology players seem to be hyping web services not only because they want to - the promise is high returns on the Holy Grail of universal data and integrated enterprise applications - but because they have to.

"Interoperability has gone from being a check-off item to a real requirement for corporate customers with a large IT infrastructure," said Jeff Mischkinsky, director of web services standards at Oracle. "That's forced the large players that weren't even willing to sit down at the same table to co-operate."

Given the high penetration and high bandwidth of the open network, proprietary and local integration has lost much of its appeal.

Yet this is hardly the first time the industry has proposed making integration a more predictable process, and just mentioning Com and Corba is enough to provoke a wince from many an IT veteran.

"It has been possible to repackage a legacy transaction as a Microsoft Com object to be consumed by a Microsoft program for years," explained Peter Havart-Simkin, senior vice president at vendor NetManage.

"More latterly, it has been possible to wrapper that same transaction, say an order transaction, from a host system as a Java Bean or an Enterprise Java Bean for consumption by J2EE development environments."

The difference is that the basic technology, transport and business process information is, in the context of web services, a homogeneous single definition, instead of the vendor-specific approaches to which Corba was susceptible, according to Wolfgang Gebhard, European product marketing manager at Tibco.

Big guns muscle in
The fact that the likes of Microsoft, Sun Microsystems and IBM are all speaking about the same subject (albeit with sometimes different concepts) is a significant development. Few have missed the chance to tie themselves to the future of web services.

According to Andrews, Microsoft is deeply interested in web services as another way to build its relevance in corporate computing.

IBM, long a mainstay of high-end business application platforms, sees a way to deliver more with more stable and extensible platforms. Hewlett Packard is very interested in the details of managing web services.

Novell, not surprisingly, is playing up its strengths in the areas of directory and authorisation, as logical skills to have onboard for web services relying on UDDI and beyond.