XML: Extremely critical or exhaustingly complex?

"The impact of XML on our daily lives and business processes is going to be felt in waves over the next 20 years. Ignoring XML now is like ignoring the World Wide Web in 1993 or ignoring client server in 1989."

Charles Abrams' views on XML may sound grandiose -- for a Gartner research director with 21 years in the industry -- but he’s not alone in his zealous support of the meta-language. Bill Gates proclaimed back in 2000 that the industry and Microsoft "should build their futures around XML".

While it seems that Microsoft has followed the advice of its glorious leader, with the .Net platform built entirely around XML’s cross-platform features, end-user organisations, while enthusiastic about the potential, are being a little more circumspect with actual deployments.

This hesitation has probably got a lot to do with the relative immaturity of the meta-language. The fragmentation in XML standards and specifications can be intimidating to the uninitiated and potentially damaging to anyone using XML for mission-critical operations.

Since XML first emerged from the World Wide Web Consortium (W3C) in 1998, the number of XML-associated standards has accelerated at a phenomenal rate. Late last year Gartner issued a report warning that unless there is a major change in the way XML-defined standards are developed, it will cost businesses dearly in the long run.

"At the very least, the proliferation of standards could result in millions of dollars in lost effort for organisations utilising the meta-language; at worst it will corrupt data and compromise business-critical transactions and operations because different parts of the same company will process conflicting XML messages without knowing it," the report concluded.

Gartner predicts that, from 2001 to 2004, companies will spend more than $3bn worldwide on XML modelling activities with no return on $2bn of it.

The problem lies with the lack of rules on how an XML standard should be developed. For example, TaXML, an XML standard relating to the lifecycle of a tax case, and Value Chain Markup Language (VCML) were originally developed by single companies. The fact that one company can create a standard is a contradiction in terms but there is nothing to stop it happening.

"There are so many specifications under development in XML right now that no one can keep track of them. There are hundreds of them," said Gartner's Abrams.

Some standards have been developed by groups of global companies such as electronic business XML -- ebXML -- which is backed by the Organisation for the advancement of Structured Information standards (Oasis) and the United Nations Centre for Trade Facilitation and Electronic Business (UN/CEFACT)-- but there are no rules or regulations laid down to enforce this approach.

But despite these efforts at coordination, Abrams claimed it is impossible for one organisation to take overall responsibility. The W3C is only concerned with core standards such as the meta-language itself, while Oasis is focused on a few key niches.

"There will never be a master international organisation handling every single XML effort under development. Some will be standards under W3C; some will be standards under the business umbrella of Oasis. Others will be vendor specifications," said Abrams.

Although the sheer number of standards seems shocking at the moment, they will reconcile over time as companies become more experienced at applying XML to their business processes, he added. Abrams said that it is up to users and customers carrying out XML implementations to make the whole system workable in the long run.

"It's not just a matter of agreeing on specifications, it's getting experience of what you do with it. Just because you have letters, grammar and an alphabet does not mean you are going to press a button and end up with the literature overnight," he said.

Given these standard issues, some experts are claiming that getting involved in a serious XML-based implementation could be more trouble than it is worth and companies should stick with tried and tested technologies such as Electronic Data Interchange (EDI) where possible.

Recent research from US management consultants McKinsey & Company suggests that there is little to gain from replacing EDI with XML in the short term. EDI has been the de facto technology for electronic commerce for more than 20 years, allowing companies to exchange purchase orders, invoices and other commercial documents over dedicated links. In 2001 more than $2tn in business transactions passed through EDI networks, according to McKinsey.

At the moment XML costs as much if not more than its EDI equivalent, XML isn't widely deployed, vendors are relatively inexperienced, and technical standards are in flux, the report claimed. McKinsey also questioned whether trading partners really want to exchange the kind of business-critical information that XML now allows for.

The overall advice seems to be that XML should be seen as complementary to EDI rather than a replacement.

"Companies can work within their industries to promote the adoption of XML standards that would extend their existing EDI infrastructures, and these hybrid systems can then be used to exchange more data with external partners," the report stated.

Gartner's Abrams agrees that comparing EDI with XML is a non-starter. "They are apples and oranges. You can take EDI right now and incorporate XML objects into it -- by wrapping the messages in Soap you have very basic Web services that can interact with other applications. Some organisations are doing this right now."

He says EDI is limited at the moment as different industries and countries use different specifications that are not interchangeable. But by integrating XML into their EDI processes, companies can make EDI more interchangeable, without having to replace legacy systems completely.

"It's going to be the work of the EDI community through XML to make these processes more interchangeable in real time with more information resources," said Abrams.

There is a well-documented tendency in the industry to vastly overestimate the short-term potential of a technology, but vastly underestimate the long-term impact.

XML is no different, according to Abrams. "If you have two different ERP applications that you want to share with business partners, you can definitely use Soap-based XML right now to start sharing data. But if you have a project with 90 trading partners and 9,000 customers, and you think you are going to press a button and have a whole new architecture for exchanging information after two months' worth of development, you’re being completely unrealistic."

Despite the relative immaturity of XML, the fragmentation of standards and the technical complexity, Abrams maintains that the meta-language is developing extremely well and users should not be put off. Indeed, it's only through companies implementing XML-based projects that the language will mature, and as expertise increases there will be a corresponding drop in costs.

"For something that was on the horizon in 1997, to go as far as it has in five or six years it has been doing incredible well. But we definitely have a way of overestimating things in the short term and under estimating in the long run. It may take the rest of the decade. You can’t expect something to take six years," he said.