InformationWeek.com

Welcome Guest. | Log In| Register | Membership Benefits

November 1, 1999

XML: The Language Of Integration
continued....page 3 of 3

Related links:

sidebar: Vendors Serve Up A Smorgasbord Of XML Support

Ardent Boosts DataStage Product Line

BizTalk Could Spur XML And E-Business

And from our sister publications:

InternetWeek XML Comes Of Age
Ultimately, XML is being touted as the long-sought lingua franca for data integration. "The main use of XML is to exchange data between internal systems, or business-to-business," says Gartner Group's Driver. By adopting XML, companies can eliminate proprietary data formats. And, unlike EDI, which also facilitated data exchange, XML is lightweight, flexible, and highly accessible. Adds Bodkin, "Data integration over the Internet is the killer application for XML."

For instance, Automotive Rentals Inc. in Mount Laurel, N.J., a global fleet-management company, maintains extensive data on the vehicles under management in multiple systems around the world. But if a customer wanted to see a consolidated report on all its vehicles worldwide, ARI had a difficult job coming up with the information. The problem resulted from information buried in multiple database systems, stored under different database schemas.

Realizing how difficult it would be to get every unit in the company to agree on a common database system, the company turned to XML, says Bill Kwelty, manager of customer services. The company was able to get everybody to agree to a set of common fields, which it defined in XML as a document type definition file. These fields represent the data customers will likely want.

With agreement on the fields and the distribution of the corresponding set of XML tags, the company, using the Bluestone XML server, is able to generate XML documents in response to customer requests for information. By using XML, the company avoided having to build a data mart or data warehouse and eliminates all the problems associated with updating and synchronizing data marts. "Now everyone has access to the information without our having to build and maintain a data warehouse," Kwelty says. All anybody needs is the ability to parse an XML document.

Callcenter Technology Inc. in Woodstock, Ga., also uses XML to pull together disparate data sources. "We can use XML to talk to the phone switch, databases, and other applications," says Todd Palmer, development engineer.

Unfortunately, most of the third-party systems with which Callcenter needs to integrate don't yet support XML, so "the benefit for us right now is small," Palmer says. In the future, however, he expects the industry to coalesce around XML and share DTDs. "With a DTD, it is very easy to bring data into our system," he adds.

The DTD is the key to the future of XML. The DTD provides the meaning for XML tags. By including the DTD file along with the XML document, any system that supports XML can read and understand the data inside the document. If you were in the automotive industry, for instance, you might define the tag

<automaker> as the name of the vehicle's manufacturer. The tag <mileage> could be the latest odometer reading. XML and the DTD don't affect how the company internally labels and stores its existing structured data. If car manufacturer names are stored in the AUTOMKR field in your database, you simply map any XML data tagged <automaker> to AUTOMKR in your system. A business partner might map <automaker> to a field named MANUFCT, but it doesn't matter. Since you both know what <automaker> means, it works.

The real power of XML, Gillett says, comes when trading partners have agreed upon a common vocabulary in the form of the DTD. By standardizing on a DTD or set of DTDs, trading partners can fire off XML documents and send and receive requests for data quickly and easily. As things change, it's easy to add new definitions, on-the-fly if necessary.

Creating a DTD, which is basically a text file, is not a big technical challenge. Developers compare the task to the effort required to define a database schema. The biggest challenge will be forging the agreement between trading partners, but even that should not be overly difficult, since nobody is being asked to change their underlying systems.

Right now hundreds of companies--including heavyweights such as Microsoft and IBM--trade groups, and ad hoc business communities are starting to put together DTDs for particular areas of interest. The problem will not be a shortage of DTDs but a surfeit. "There will be competing DTDs," Driver says; but he's not bothered by it. Through a process of natural selection, preferred DTDs will rise to the top.

With the arrival of DTDs, using XML should be a snap. Unlike learning Java programming, mastering XML is pretty simple, according to developers. "It is like learning HTML. There is some structure, but it is quite easy," says Palmer, who learned XML out of a book.

While HTML provides a great starting point, says Bodkin, there are some data design tricks to master. It is possible to write a poor DTD, one in which the tags are obscure, unnecessarily cryptic, or inappropriate. "XML is a data-oriented language," Roche says, "so it's not difficult for data modelers."

Given the advantages of XML and with no great hurdles to overcome, an easy learning curve, the rapid proliferation of DTDs, and support for XML cropping up in a wide range of products, XML appears as close to a sure bet as any new technology developers are likely to encounter. Says Lamoureux, "XML is just too compelling to ignore."

return to page 1, 2