Funny thing about XML database servers: Even though XML is by now firmly established as the lingua franca of the next phase of e-commerce, conventional wisdom has it that dedicated XML databases are not yet ready for prime-time enterprise use. With no robust XML query standards in place and few examples of large XML database deployments to study, most IT managers have favored the more familiar relational databases for storing XML content.
Conventional wisdom, however, has not stopped the U.S. Air Force from putting an XML database server at the center of a new global, Web-based document access system. Already, that decision has begun to pay off, allowing one major Air Force weapons system to save more than $800,000 to date. Its also established that XML databases can scale, so much so that the Air Force is now significantly expanding use of XML database technology.
The test case for XML database technology in the Air Force was a project, launched early last year, to provide searchable Web access to some 150,000 pages of technical documentation related to the E-3 AWACS (Airborne Warning and Control System) surveillance aircraft.
Before the project, thousands of pages of AWACS documentation had been stored as SGML (Standard Generalized Markup Language) files, printed and distributed to maintenance, procurement and other Air Force officials. The process could take 45 days for printed material to reach users. In search of a way to cut document update and distribution costs, Air Force officials with the U.S. Air Force Electronic Systems Center at Hanscom Air Force Base, in Bedford, Mass., came up with a plan to translate the SGML files to XML, store them in a central repository and provide searchable, browser-based access to them.
Air Force officials contracted the project to Veridian Corp., an integrator based in Arlington, Va. Veridian designed a system composed of two parts, both implemented as Web services and both interfacing with an XML database: TextML Server from Ixiasoft Inc., of Montreal.
One part of the system, called TPMS (Transition Process Manager Services), mainly validates and translates the Air Forces SGML documents into XML using the OmniMark tool from OmniMark Technologies Corp., of Ottawa. Users at four Air Force locations responsible for maintaining the AWACS documentation entered the SGML documents for translation using a publishing client that interfaced with the TPMS system through a Simple Object Access Protocol API.
The second part of the AWACS system, called EDMS (Enterprise Document Management Services), lets Air Force personnel at bases around the world query the XML database and view documents via Web browsers and the Air Forces private network. Like the TPMS portion, it uses Web services APIs to access the TextML XML database.
EDMS also allows users to query the AWACS XML documents using a proprietary query engine built into the TextML database server product. Users can query multiple documents at the same time by dynamically mining XML tag information—part number or subsystem name, for example—that is used to create a table of contents showing whats in a given document or set of documents.
Air Force and Veridian officials said theyve been happy with the decision to build the AWACS system around the Ixiasoft TextML XML database rather than a relational database. Using a relational database to store and provide access to the AWACS XML data, said Bruce Watson, Veridians project manager, would have been much more expensive, requiring his team to write code to transform standard XML into a form that could be stored in rows and columns, then transform it back again for viewing.
But Watson and Air Force officials acknowledge that relying on relatively immature XML database technology has required trade-offs. In using the TextML XML database server, for example, Veridian and the Air Force had to buy into a proprietary query engine. The query approach used by Ixiasoft does not make use of the so-called XQuery proposed standard recently updated by the World Wide Web Consortium.
Until Ixiasoft brings its product in line with XQuery, the Air Forces AWACS system will rely on one vendor for its query function. That means it could be difficult for the Air Force to disentangle itself from the technology were Ixiasoft to go belly up or were a better product to come onto the market.
Watson said he and Air Force officials were fully aware of the risks. But, he said, they were minimal in light of the potential reward. Ixiasoft officials said the company will support XQuery once it becomes a fully accepted W3C standard. In the meantime, TextML stores data in standard, native XML. So, at least in theory, the Air Force could switch XML databases or use a different, standard query engine if necessary.
“We realize were running out on the edge a little bit here, but we feel were not leaning out over it,” Watson said.
Nor do Air Force officials, apparently. Now that the AWACS system has been deployed, the Air Force is considering using the technology developed by Veridian to access documents related to other major weapons systems. Material Systems Group officials at Wright-Patterson Air Force Base, near Dayton, Ohio, will soon begin to test the system, which has been renamed Air Force Common Viewer. If tests prove the system can handle a large volume of documents—potentially 16 million pages, said Common Viewer Project Manager Steven Holloway—deployment of the Common Viewer system throughout the Air Force could begin in a year.
“This is all about improving efficiency and cutting costs,” Holloway said.
