The XML Approach

The metadata and shared context concepts are general, and there are many different ways to implement them. XML is an increasingly popular implementation, and it will almost certainly be the most widely used in the near future. XML is an outgrowth of Standard Generalized Markup Language (SGML), which became a standard of the International Organization for Standardization (ISO) in 1986. SGML had its origins in IBM, which wanted a means of describing document content so that it could publish the same content in a number of different ways. For instance, it would be nice to be able to deliver the same technical information about a new product feature both as a stand-alone bulletin for existing users and as part of a user 's guide for new users. Because other organizations were struggling with similar problems, it made sense to create a standard for document markup. The result of the standards process was a rich document markup language that allowed authors to separate the logical content of a document from its presentation. The fundamental approach was to add metadata to denote this logical structure and provide shared context for specifying the metadata rules followed by each type of document.

HTML is also a descendant of SGML, but it was never intended as a general means of structuring information exchanges. It predefines a narrow set of metadata specific to online publishing. SGML and HTML leave the requirement for a standard means of defining metadata on the Internet unfulfilled. However, when people began to develop such a standard, they naturally looked to SGML as a starting point. The World Wide Web Consortium (W3C) formed a working group to study the issue in July 1996. This group 's goal was to produce a simplified subset of SGML suitable for use on the Web. Why a simplified version? SGML is extremely complex and poses some problems for automated processing of large volumes of Internet documents. It seemed that a subset of SGML could be simple enough for people to understand in its entirety, yet expressive enough to meet the need for shared context on the Internet. The result was the XML 1.0 specification, released as a Recommendation, the highest level of W3C endorsement, on February 10, 1998.

The XML approach to metadata and shared context is simple. Authors add metadata through tags . The syntax for adding these tags is very similar to that of HTML. To indicate that William Shakespeare is the author of a work, one could write "<author>William Shakespeare</author>." Document designers add shared context through document type definitions (DTDs) . A DTD specifies the allowable order, structure, and attributes of tags for a particular type of document. A document references the DTD that governs its structure with an Internet uniform resource locator (URL). Therefore, any party can spontaneously access the DTD, interpret its rules, and process the document.

While DTDs are useful, they are insufficiently precise in certain cases. The more sophisticated the application, the more precise the shared context must be to avoid misunderstandings. In many cases, the level of precision must match that of the programming languages and databases used to process the information. A new W3C standard for XML Schema addresses the need for more precise shared context and will likely supplant DTDs for many applications. It became a Recommendation on May 2, 2001.

While available for only a few years , XML has experienced rapid adoption. Because of the convergence of information exchange problems, the application of XML in one area has led to the application of XML in other areas. As a critical mass of solutions has emerged, much of the software and networking infrastructures have developed direct support for XML, including the following.

• Development tools. To assist programmers in creating application code that works with XML, development tools have evolved XML-specific features that improve productivity. These features go beyond the simple inclusion of low-level software components for processing XML data. They include utilities for generating XML processing code, manipulating information extracted from XML documents, and distributing this information through different channels.

• Application servers. Because XML facilitates information exchange in distributed systems, organizations often deploy XML-based solutions on three-tiered architectures, with an application server performing much of the processing. Therefore, application servers have evolved features for processing XML-like frameworks for user presentation, interfacing between business documents and business objects, and assembling documents from database data.

• Databases. Clearly, organizations need to store mission-critical XML data in an industrial-strength database. While developers can use the application layer to assemble and disassemble XML documents, it is even more convenient for the database to handle them directly. Toward this end, databases are evolving features for mapping documents to tables, directly storing and searching XML data, and performing sophisticated operations on this data without ever taking it out of the database.

• Networking software. Because many XML-based solutions involve sending and receiving documents across networks, developers need networking software that supports such operations. A family of emerging distributed XML messaging standards has provided a starting point that ensures compatibility among such software. This seed has led to the widespread availability of components that implement these standards and provide high-level interfaces to functions critical to the development of distributed XML applications.

• Networking hardware. With the potential explosion of important data flowing over the Internet in XML format, there is the opportunity to improve the quality of service by putting XML support directly in the network hardware. The self-describing nature of XML makes it possible for this hardware to take actions based on the message content, so routers can expedite the flow of important messages and even determine their eventual destination based on rules describing which machines can process which types of XML information.

This widespread support for XML has created a network effect similar to the one that catapulted the Web to prominence. It seems as if everyone is using XML for some purpose. In some sense, the W3C greatly exceeded expectations. It developed a technology that not only solved the information exchange problem on the Web but has the potential to solve information exchange problems in every area of distributed computing. Unfortunately, the generality of the solution poses a problem. Because the potential applications are so broad and the support from computing infrastructure is so widespread, there is the opportunity for confusion. Organizations have a hard time sifting through all the possibilities to identify how XML can best help them and then selecting the XML-based technologies that best match their situation. It is therefore important for managers charged with solving information exchange problems to have a solid understanding of how XML works, how to identify opportunities for its use, and how to overcome the inevitable barriers to deployment.