Enterprise Apps: XML Takes On E-Commerce

When Extensible Markup Language was approved by the World Wide Web Consortium in February 1998, the standards gurus hoped XML would offer a more efficient way to publish Web pages. But developers quickly realized that the power of defining their own tags separate from the file contents meant data could be defined and easily exchanged. This ability, tied tightly to Web publishing initially, spread to more data interchange situations and now has XML taking on e-commerce.

Companies trading together must share common data definitions, such as whether address means a physical address or a network address. XML syntax uses matched start and end tags, such as "<address> and </address>" to define information within a file. This syntax has three advantages: it's easy for humans to read, easy for machines to process and has the look and feel of HTML.

"XML makes quick and dirty implementations compared to EDI," says Benoit Lheureux, research director for Gartner Group's application integration and middleware strategies division.

EDI is the e-commerce standard used by some large companies and their trading partners, but is considered too complex and overkill for medium and small companies. Unlike XML, EDI is not text-based and is hard for people to read. "XML makes it easy for vendors to agree on [formats] for purchase orders and the like,'' Lheureux says.

Developers and tools vendors are quickly exploring XML, and products using XML are readily available for Web publishing applications. Many data integration and exchange products powered by XML have been announced, but most won't ship until this summer.

"XML allows us to pull from any data source, it simplifies data distribution and optimizes data exchange,'' says John Zollinger of Arkona Inc., marketer of Universal Update, a middleware product that uses XML. "Once we finish the adapter to a legacy data source, the resulting XML-enabled data is simple to change and update."

One of Arkona's customers is a trucking company that's implementing the Universal Update server to run EDI-like transactions with 3,000 customers.

Using any report writer or Perl script, the trucker's customers post scheduling needs on a Web site checked by the Arkona server software. Schedule changes are noticed and the trucking company's schedules are changed to reflect the customer request.

Again, no changes are required to existing databases, just the conversion routines from the report output to XML for the Arkona server to check for new instructions. The customers need do nothing after the XML software installation, unlike using an EDI system, where they must implement and coordinate EDI transactions with the trucking company on a regular basis.

"XML itself is just a vehicle to define what's in a source or describe what's in a document," says Andrew White, vice president of development and marketing for Sagemaker Inc., which developed an XML-based middleware solution that lets it deliver data from 4,000 different sources to the petroleum and chemical industries.

"We use [XML] tags to organize that information into a common classification system,'' he says. Data sources vary widely, including drilling records, balance sheets, news sources and even maps.

Sagemaker developed a proprietary version of data tags for internal use before XML was released, so searches across data sources could use a set of common keywords. New data are now tagged using XML, and earlier tags will be replaced by XML over the next year and a half.

"XML is like an open EDI,'' White says. The power of XML, he explains, is "that people are agreeing to a common technology and common [formats], especially in specific vertical markets."

Examples cited by White include his own energy industry, along with the financial, chemical and pharmaceutical industries. Companies now use XML to send product catalogs to customers and to conduct real business, including exchanging purchase orders, receiving acknowledgements and sending bills.

Other services build on the idea of hiding the XML plumbing from customers. Celarix Inc. greases the wheels of supply-chain commerce by acting as an intermediary for companies. Using the outsourcing model, Celarix removes supply-chain transaction applications from companies to its own hosting facilities. Supply-chain partners use browser clients to manage transactions instead of using EDI systems in each company.

"We use a series of standardized XML DTDs," says Graham Lubie, chief development officer for Celarix, referring to the Document Type Definition file, specifies the names used for data element types, where they may occur in the document and how they all work together.

Using Enterprise eCommerce software from IPNet Solutions Inc. for transport, Celarix receives information from clients, processes that information on its own servers and provides the result to customers of the original company. Celarix converts client data into XML, then does the same for its client's trading partners. Neither client nor partner need change their internal systems, yet all benefit from the use of XML for data connections. Easily understandable XML tags and processes replace the more complex EDI definitions and transactions.

One company racing to join the XML fray is Sterling Commerce Inc. with its Gentran electronic-messaging server. Working with Ariba Technologies Inc., Sterling joined more than 40 other companies supporting the Commerce XML (cXML) initiative. Dave Wilkinson, director of product architecture for Sterling, says that what's missing in XML are real schemas and document-level semantics. The cXML group is one of the first to try to fit the document schema into the data-flow environment, he says.

XML is certainly driving e-commerce, but it's also useful as a way to speed interchange between disparate internal databases. One good example is Dun & Bradstreet Inc., which tracks credit information on more than 51 million companies. Unfortunately, the databases holding the information vary widely between regions in the United States and countries around the world. As the company grew by acquisition, it inherited at least one of every database hardware and software combination imaginable. XML, in its role as universal data interchange format, serves as a master translator between the different databases.

Tom Gwydir, project director for the Global Technology Organization within Dun & Bradstreet, used webMethods B2B to develop a single global-access system with a common data interface and access over the Internet. Clients with a browser must have easy access to data from every database, regardless of location or original data format.

Previously, custom client applications had to be designed for each data source needed by the client, raising costs and slowing implementation. Gwydir standardized on XML and webMethods in June of last year, and released version one of Dun & Bradstreet's Global Access Toolkit last December. Three customers are in production already, another dozen are close and yet another dozen have started the process.

"By the end of the year, we should have a fairly large number involved," Gwydir says. "Previously, it took three to six months to devise proprietary data packets and the dedicated link for companies to use our data. Customers can now download the toolkit one day and go into production the next, thanks to XML.''

Working Out The Kinks
The XML/EDI Group is working to ensure XML and EDI fit together smoothly. EDI proponents already built a framework with an element dictionary and messaging format descriptions. XML should develop more quickly by taking advantage of that history while avoiding the mistakes of making complicated and expensive systems for conducting business. According to the XML/EDI Group's own Web site, only 2 percent of American businesses have registered for EDI, which means that there's a real need for an easy-to-use data integration standard for e-commerce.

Most vendors and users in the XML space agree that more business-to-business cooperation is necessary to make XML an e-commerce success. Specifically, companies in the same markets must agree on content definitions.

"People are less interested in features than in portability and interoperability," says Winston Chung, director of middleware and Web integration for American Management Systems Inc. Although few XML applications have shipped, Chung understands why people are excited.

"XML offers simplicity,'' he says. "Before, when companies exchanged information from one end to the other, there was a one-to-one negotiation before agreeing on bits and bytes. XML gets away from all the lower-level detail, effectively using text with special tags and a way to define the meaning of those tags."

Companies that don't board the XML train will be left in the dust, according to Chung and others. "If you don't talk XML and use popular tags, your program won't interoperate with others, and that's pretty scary," says Chung.

"If your data doesn't follow the search tools and keyword formats others use, you will not be found,'' says Sagemaker's White. "When 40 percent of an industry works together, they will tip the rest into compliance. It becomes almost Darwinian: conform or die."

Only the computer industry would complain about shortcomings of a technology not yet fully implemented, but there are gaps in XML. Already, many observers believe the DTD specifications should be upgraded.

Document Content Description (DCD) is a superset of DTD rules that includes additional properties, such as basic data types. Since data extraction is a common job for XML, DCD includes database interface features to better handle concepts such as key fields and the uniqueness of certain values. The DCD framework was developed by a consortium that consists of IBM, Microsoft and Textuality and is under review by the W3C.

White says the expectations for XML are the biggest problems. "This isn't rocket science; it's a bit of a yawn as a technology," he says. "The key thing is doing with open standards what everyone used to do with proprietary software."

Legacy systems must be replaced by XML-enabled data controls, or filtered to present an XML interface to the world. Many industries are moving toward XML-tag consensus. Both the chemical industry and academic mathematical groups have defined common tags. Some tags, however, are more complex. "I can easily give consumers temperature readings at five international airports," White says. "But I can't give you five comparable profit figures for international companies, because profit means different things in different countries."

Chung chafes at the lack of common tags. "The conversation between employers and 401(k) insurers have no defined processes and tags,'' he says. "There is no common standard for communicating payroll deduction information, for instance."

Gartner's Lheureux sees XML's good and bad sides. "XML is simplistic and opportunistic, but doesn't enforce more rigorous agreements like EDI does."

Zollinger from Arkona sees the problem as political, not technical. "XML is a subset of SGML," says Zollinger, referring to the Standard Generalized Markup Language standard developed in the 1980s. "There are lots of guidelines available we can apply to XML, but many companies are still reluctant to share information with other companies they don't know well."

Sterling's Wilkinson welcomes those who are afraid to share XML openly. "We perform translations from XML at one customer to XML at another, so we do well no matter what," he says. "But lots of industry groups are already defining the XML translations and tags for their member companies."

For XML to really take on e-commerce, vendors still need to pack security features like encryption and the ability to handle digital certificates into products. And users say enhanced compression is needed. But XML is well beyond its original status as a Web publishing tool. The bottom line is that this technology is not hype. XML just may be the catalyst that will extend e-commerce into the mainstream.

James E. Gaskin is a technology writer and consultant based in the Dallas area. He can be reached at james@gaskin.com.