XML databases gaining acceptance

The surge in XML data – and in Web services to access it – are forcing companies to create new information architectures with XML data stores as the key component.

Just when customers expect native XML database products to come into their own, however, those products are facing growing competition from traditional database vendors Microsoft and IBM. In addition, Oracle and Sybase are working to persuade enterprise IT groups to adopt these next-generation data stores, along with data-management and application-development tools, as the building blocks of a new enterprise information architecture.

The use of an XML database often originates in a specific project. Command Financial Press, a New York publisher of financial information, uses the Ixiasoft XML server to store and manage content for the prospectuses its mutual fund clients – each of whom may have scores of funds – publish yearly. “Much of the data is unique, but a lot of it is common to all of a client’s funds,” says Will Montgomery, director of project management with Command’s IT group.

In the past, customers had to treat each document as a separate entity to be written, proofread, modified, proofread again and so on. Even boilerplate wording – otherwise identical – might have to be changed to identify each fund by name. “If the customer had a hundred funds, they had to make a hundred separate changes every time,” Montgomery says.

Now a client logs on to the XML-based system through a secure Web site and works with the documents in Microsoft Word 2003. The XML server stores all text as components, which it assembles into finished documents on request. Changes are made once, then replicated through related documents as needed.

Command Financial is evaluating the idea of applying this same XML infrastructure to the unstructured information in shareholder reports.

Enterprise XML adoption

In a recent study by IDC (see graphic), about 29% of about 500 corporate IT respondents said they are widely using XML content repositories and databases. Almost the same percentage said they are exploring such use. In addition, the study found wide use of XML technologies, including editors, XML-based electronic forms and XML schemas. Each technology is widely used by roughly a third of study respondents; almost exactly the same percentage of respondents said they are exploring the use of these technologies. This rising interest has been one factor in traditional database vendors’ product plans and in the ambitions of native XML-server vendors.

Microsoft late last year released SQL Server 2005, code-named Yukon, which can store and process XML data without having to convert it into relational rows and columns or store it as a binary large object. Programmers can query the XML data using XML Query or XQuery, a language nearing final approval by the World Wide Web Consortium (W3C).

IBM is beta-testing DB2 Viper, which will be able to store traditional relational data and XML data natively. Viper is due out later this year. IBM officials have been relating Viper’s XML data-management strengths explicitly to the requirements of service-oriented architectures, where programs and data in all formats can be categorized, found, accessed and used via standard Web-services interfaces.

The W3C is in the last stages of creating a final recommendation for XQuery, which will create a standard query language for accessing and processing stored XML data. It will be the XML equivalent of the SQL language for relational databases and vastly simplify programming of XML applications.

XML is increasing the common representation for a growing chunk of information that is unstructured: documents, reports and forms. “High-end publishing applications, for things like technical manuals in aerospace or automotive, have been using XML for a while,” says Rita Knox, Gartner vice president of the high-performance workplace group. “But now it’s starting to move into near areas, such as banking. A common [XML] representation called the Extensible Business Reporting Language is being developed in banking, for sending reports on assets and loans and other information to the [Federal Deposit Insurance Corp.].”

What can I do with it?

The key question, says Melissa Webster, program director for content technologies at IDC, is: “What do you do with this [XML content] in terms of providing content services?” The answer is driving the next phase of development, she says.

In general, she says, native XML database products and the emerging XML capabilities in traditional databases do a good job with the basics: scaling to handle large stores, good performance, managing versions of an XML document, linking pieces of content.

The real payoff, Webster says, comes in two more-advanced areas. One is revising content continually in a way that marries, say, updates or revisions to a technical manual with annotations and notes created by engineers using it in the field. Webster calls this configuration management.

The second, more important payoff is linking stored XML information with critical business processes, such as handling a mortgage loan or making a repair to a jetliner. You could start with aircraft CAD engineering drawings, generate from them finished technical documentation for engine-repair mechanics, then link a repair ticket to a specific subset of instructions and drawings, and feed back workflow milestones into the repair history of the engine and to reports that are generated for the manufacturer and the Federal Aviation Administration.

“In the past, technical manuals lived in silos separate from the business processes that made use of them,” Webster says. “The humans, the mechanics in this case, had to be the go-betweens. Intelligent XML content-serving lets you merge these together, business processes with specific content.”

This potential is fueling the ambitions of vendors of native XML products. Despite looming competition from companies such as IBM, Microsoft and Oracle, investors are backing new start-ups such as Mark Logic, a San Mateo, Calif., company that offers an XML content server.

“If XML content is simply data that’s wrapped in XML, there’s no reason not to use Oracle or Microsoft,” says Max Schireson, vice president of customer solutions for Mark Logic.

Intelligently managing text and other content in complex documents and processes is the kind of problem relational databases don’t handle well, however. Publisher O’Reilly Media uses the Mark Logic server to create a system college professors can use to create a custom reader for a given course. The professors can do complex searches of the technical content from O’Reilly’s vast library of books and publications, stored as XML documents. They add their own content and place an order to have these selections printed on demand, bound and shipped to their office.