March 5, 2001

http://www.informationweek.com/827/xml.htm

XML-Like The Air That We Breathe

XML will change the appearance and substance of corporate systems. A big part of the language's appeal is in being able to meld business processes with application

By Martin Marshall   (mmarshall@zonaresearch.com)

ML is poised to affect just about everything corporate IS does, from E-commerce applications to legacy data. But the pervasive changes XML will bring about won't become apparent until the XML products under development hit the market later this year.

IT managers expect XML to fundamentally improve the speed, cost, and flexibility of their business applications. It's also expected to alter the way they build new applications and integrate data from existing systems.

XML will have a profound effect on business processes, easing the task of exchanging data with trading partners. To some, XML is a business-process catalyst that will pick up where electronic data interchange and extended intranets fell short.

Zona Research predicted early last year that the percentage of E-commerce transactions using XML would rise from .5% in early 2000 to more than 40% by the end of 2003. In a Zona Research Market Report, "XML: The Dash For Dot.com Interoperability," released last month, a survey of more than 200 companies indicates that IT managers expect XML to dramatically improve the adaptability of their businesses.

XML is much more than a markup language; it's a fundamental mechanism for the automated exchange of data and the processes that act on that data. XML's data-transformation mechanisms go beyond operating environments, transport protocols, and the arcane barriers of the applications to present true interapplication communication.

XML covers everything from data and data-transformation processes to schema, development tools, XML servers, and components. XML also takes into account business-process mechanisms, layered architectures, and vertical-industry bodies that make decisions about XML data representations and process definitions for their industries.

XML could supplant EDI as a mechanism for transferring data between businesses and their applications. EDI has been the main way that companies exchange business forms. EDI transactions total about $750 billion per year, with about $2 billion a year spent on EDI development and deployment, according to Zona Research.

EDI does the job for bidirectional interaction, but it's expensive to implement, and the embedded business rules are rigid. EDI is a point-to-point solution that must be reengineered every time a company adds a business partner. The mapping of data sets and procedures between two trading partners in an EDI environment is generally accomplished by custom coding.

There's a growing movement toward converting EDI systems to XML, according to Zona Research's survey. Among the 72% of respondents who use EDI at their companies, seven out of eight plan to convert EDI into XML at some point. The largest group, 30%, will convert some of their EDI to XML this year, while 14% will do some conversion next year or later. They'll do it on a selective basis, however; very few will convert all of their EDI to XML by either 2001 (2%) or 2002 or later (4%). About one in eight will convert EDI to XML on an as-needed basis.

XML Solutions Corp. is an early implementer in converting EDI to XML. Its XEDI product claims to be able to handle all of the ANSI X.12 EDI interfaces.

With its many technical twists, it's easy to overlook the political movement behind XML. As such, it's not born of rosy optimism about global cooperation, but rather about the expedience of operating in trading communities rather than as closed systems.

Each vertical industry has a major XML effort under way to define the data term definitions and schemas for industrywide exchange of data. A data term definition is much like a database mapping tool, where a field in one table is mapped into the equivalent field in another table. The collection of data term definitions that completes the form conversion is a schema.

These efforts involve not only term definitions, such as the fields on a health-care insurance claim form, but also the processes that operate on this data, such as the calculations that are applied to a definition of a 30-year mortgage in the financing industry.

The health-care industry's initiative is being conducted by the XML subcommittee of the HL7 group, which in October adopted a new version of the Clinical Document Architecture that uses XML messaging and data structure.

There are two initiatives going on in the electronics component industry: RosettaNet and the United Nations-led ebXML effort. In insurance, it's Acord's schema efforts. The publishing industry, represented by IPTC, has achieved unification behind NewsML. The mortgage business has developed schemas using fpML; aecXML and LandXML in the architectural industry; the Retail Business Information Schemas in retailing; and similar efforts in petrochemical, real estate, food service, and manufacturing.

XML implementations aren't far enough along the experience curve to provide more than anecdotal evidence of business-process acceleration, but some claims have surfaced, many of them coming from RosettaNet members.

Lucent Technologies Inc. says XML-based technical information dissemination has cut its engineers' component selection time in half and reduced risk-assessment time by 75%.

Arrow Electronics Inc. has implemented nine of RosettaNet's Partner Interface Processes with 12 suppliers and one customer. The turnaround time to the customer has been reduced from the next day to the same day, cutting 15 hours off a process that was previously handled by EDI.

The health-care industry is also moving from EDI toward XML-based processing, albeit more slowly than RosettaNet and the electronics-component industry. Part of the rationale for these standards initiatives lies in the human readability of XML-tagged health-care records, compared with the indirect readability of EDI-encoded records, which are untagged fields of data separated by vertical bars.

The Mayo Clinic's Clinical Notes Project report was instrumental in swaying health-care institutions to vote for HL7's XML-based version 3 messaging, but health-care officials are reluctant to make statements about quantifying any perceived business process or error-rate improvements. But it's agreed that new XML-based applications will be used across the health-care industry, rather than being confined to proprietary formats on an institution-by-institution basis.

Business IT executives usually have strong technical reasons for making strategic investments, such as those that would be required for XML-related projects. Many IT managers indicate plans to convert existing data into XML form to enable enterprise search capabilities. Users have long felt frustrated by not being able to access, let alone search across, data that they know exists in their company, but which isn't in a searchable form.

IT managers expect XML technology to shorten application-development time and enable the performance of new operations once the data is in XML form. It's too soon to obtain broad quantitative data on how much application-development time will be saved, but the few IT managers who were willing to make an estimate considered themselves conservative in estimating 30% to 50% speedups in development time.

XML means that programmers will have only one data interface to program instead of many, and they'll no longer have to embed all business processes in each application. XML will let business processes be shared across applications.

The price of admission is that programmers will have to learn the tagging syntaxes of XML and its various processes, such as the creation of style sheets in the Extensible Stylesheet Language. The learning curve shouldn't be too steep--XML is similar to learning a scripting language such as JavaScript, rather than a third-generation language such as C++ or Java.

Only a small percentage of IT managers feel compelled to adopt XML to participate in online exchanges. This suggests that, despite the proliferation of more than 1,000 business-to-business hubs in the past year, they're not yet a major force driving business IT decisions. It also reveals the nascent stage of XML usage in B-to-B hubs.

XML has many hurdles to widespread adoption. The top concerns are security, standards development, schema development, the maturity of XML tools, and top management's lack of understanding of XML.

The security concerns are already being addressed. Microsoft, VeriSign, and webMethods published an XML Key Management Specification with the prospect of an uncontested adoption path through the World Wide Web Consortium standards body. XML encryption, which should be standardized by year's end, will enable the encryption of fields within a document, such as a credit-card number.

An XML standard for enterprise management of access-control lists and authorization mechanisms is also needed. Access-control lists are data tables that contain the information about which users can access a system, which applications they can access, and what data they're entitled to see. The authorization mechanism contains passwords and user IDs. The classic problem found in most companies is that multiple access-control lists exist, tied to different applications within a company.

There are two conflicting standards in this area: S2ML, backed by Netegrity, Oracle, Sun Microsystems, VeriSign, and webMethods; and AuthXML, endorsed by BMC, Deloitte & Touche, Novell, and Securant Technologies. Until the S2ML/AuthXML issue is resolved, there won't be industry agreement on XML-based security, so we may see early adopters promising to change their software if a standard arises.

IT managers are concerned about the immaturity of standards for embedding business processes into applications. While business processes can be embedded as tags in the XML data, in heterogeneous environments it's unclear whose XML server will control the acknowledgments, workflow, and other processing that may accompany the data.

For example, it's unclear whether Microsoft's BizTalk Orchestration implementation will be able to exchange complex business-process objects with non-Microsoft servers. The potential solution lies in agreements among IBM, Microsoft, and others over the Web Services Description Language. This is built on top of the Simple Object Access Protocol, used for remote invocation of processes between XML servers.

Schema standardization is the great boon and bane of XML. As vertical industries achieve standardization on procedures and data term definition schemas, XML servers will provide a mechanism for the commonality of data exchange on an industrywide basis.

It will be a challenge within vertical industries to agree on the intricate procedures that are most germane to their businesses. For example, the petrochemical industry must embed procedures specified by the Interstate Commerce Commission for handling hazardous materials, and the pharmaceutical industry must embed procedures specified by the Food and Drug Administration.

As much as HTML and the Web attracted businesses to exploit Internet technology for commercial purposes, XML advances will have a profound effect on automated B-to-B and internal company interactions.

---

Martin Marshall is Zona Research managing director. Reach him at mmarshall@zonaresearch.com.

---