Sticking with XML: Tying Disparate Systems Together - Technology Information

ENT, Feb 23, 2000 by Mark Mcfadden

In a forest products company in Washington state, a newly designed e-commerce site allows businesses to order pulp and paper products directly over the Web. Businesses order newsprint and other goods on the site to the tune of nearly $100,000 a day. The orders are collected on a database at the Web site and then carefully transferred to the company's existing back office system.

But what happens if the person in that all important swivel chair accidentally drops an order or two while transferring information from the new electronic commerce system to the legacy systems that supports the company? Whether it's information exchanged between partners or simply the transfer of data between newly Internet-enabled applications and older applications, something has to act as the glue between the two.

That glue is turning out to be XML.

The End of HTML

To understand how business-to-business e-commerce has been transformed by XML, it's important to understand why its ubiquitous predecessor, HTML, wasn't up to the task.

HTML is one of the engines that has driven the electronic commerce revolution. Despite this, even its inventors recognized its limitations. HTML is, effectively, a presentation tool that tells a browser how to arrange text, graphics, and links on a screen. One of the most important things missing is a mechanism for describing the data being presented.

As a result, there is no way to effectively gather data from a Web page. An incredibly patient person might cut and paste the contents of a screenful of information, but no normal person would do it on a regular basis. Suppose you asked to see a list of books on e-commerce from Amazon.com, ordered from cheapest to most expensive. Taking that same data and resorting by the length of the book requires an entirely new interaction between the server and the browser -- generating the exact same query with one minor modification.

One solution is to give the client more information about the data by adding tags. HTML has tags like [less than]B[greater than] and [less than]P[greater than] to indicate how a browser should display the content of a page. Why not add tags like [less than]AUTHOR[greater than], [less than]PRICE[greater than], or [less than]PAGECOUNT[greater than] to let the browser recognize the data contained in the page? This strategy would allow the browser to recognize the data components of the document and make it possible to manipulate or export that data as requested by the user. Unfortunately, adding tags to HTML was one of the most difficult and time-consuming tasks facing Web standards organizations in the 1990s.

Linking Data and Documents

Instead of adding complexity to HTML, the World Wide Web Consortium (W3 C, www.w3.org) went back to its roots. HTML is based on Standard Generalized Markup Language (SGML), a complex system for describing other languages. Incredibly powerful, SGML has far too much overhead for use on the Internet. Something was needed with the flexibility of SGML, but with the lightweight features of HTML. In February 1998, the W3C produced XML.

While XML can be used to format documents in the same way that HTML does, its true power is in information exchange. XML can support structured documents that contain a view of data, such as the result of a query, a request to download a section of a database, or a dictionary entry describing information available from a Web server.

XML provides a standard, vendor-neutral way for both data and accompanying metadata to move over a network. This makes it possible for two companies to exchange information without having to understand anything about the underlying databases or information sources. To communicate, a source system simply reformats its proprietary information as an XML-compliant document. The system sends that document to any other system or application that can understand XML.

The Tie that Binds

"It's very clear that XML will be the lingua franca for business-to-business electronic commerce," says Zack Urlocker, vice president of marketing at Active Software Inc. (www.activesw.com), a developer of middleware that lets legacy applications work with newer business-to-business e-commerce solutions. "Today, customers expect all applications to work together in real-time -- yet older applications need ways to become a part of an integrated electronic commerce solution. A customer or business partner doesn't care that one part of the application is built out of e-commerce tools and another uses SAP. They just want it all to work -- now! XML is going to be an enabling tool that allows us to make the new generation of e-commerce applications and our older, line-of-business applications seem like one system."

In business-to-business electronic commerce systems, organizations can move data between companies without having to invent custom, special-purpose applications at both ends. There is also no need to invent new mechanisms to move the data around: most data will travel between companies using HTTP and message queuing technologies, such as Microsoft Corp.'s (www.microsoft.com) Message Queue Server and IBM Corp.'s (www.ibm.com) MQSeries.