If today’s XML data usage trend were made into a science-fiction movie, XML could star as the new “Blob.” But instead of aiming to destroy a large city, this XML blob aims to engulf the world’s corporate networks and smother everything in its path.

While reality doesn’t make for such dramatic cinema, the XML creep into the enterprise can’t be dismissed. In response, CIOs need to know how to deal with this new type of data so that handling it and carrying it over corporate networks doesn’t result in a significant drop in performance.

The reasons for special treatment
Two aspects of XML data are prompting tech managers to consider giving XML data special treatment.

The first is that the amount of XML data is about to increase rapidly. Three broad business application areas are driving XML usage, according to Gartner research: enterprise application integration, the interchange of data with partners (what some call business-to-business interchange), and Web services.

All three of these areas are growing in popularity, and all three make heavy use of XML data. Gartner predicts that the amount of XML data in corporations will grow from about 2 percent in 2000 to 60 percent by 2004.

The second aspect is that data stored in XML format is typically larger in pure bytes than the same data stored in other database formats, such as flat files or a relational database management system (RDBMS) format. That’s because XML data typically includes both the raw data and metadata, which describes the data.

The difference in file sizes can be quite significant, according to experts. Web services consulting firm ZapThink LLC says that 1 GB of traditional database information could expand to as much as 20 GB when all of the XML coding and descriptors are added.

“With XML, no matter how small the [payload data], you have fixed overhead,” explained Charlie McCallan, VP of product management at BoostWorks Inc., a Web acceleration software vendor. “A message with a 5-byte payload might have 300 bytes of [XML] description,” said McCallan.

The XML impact
The growing use of XML in corporations and the resulting increase in data pose significant consequences for an organization’s infrastructure.

“Because of XML’s verbose nature, the handling of XML data will strain application servers,” explained Danny Briere, CEO of the consultancy TeleChoice Inc.

Since XML data often needs to be extracted, parsed, and frequently transformed into other formats like HTML, it can cause a significant drain on CPU cycles from an applications server.

That was the case at Hemscott LLC, a UK financial information business whose customers include European financial service firms, public companies, and investors. The company provides customized business research through a financial portal that gets over 13 million Web page views each month.

The site data is stored in XML format and needs to be converted into HTML on the fly, depending on each visitor’s query. The organization went with XML as the database format since it wanted to use the same data for more than one purpose and present the data in a wide range of formats.

That said, the large XML file sizes (about 500 KB per file) meant that the conversion process required to display the data through a client’s browser took considerable time. This slow site load wasn’t acceptable to Hemscott’s customers or its level of customer service.

“We provide a high-end service to demanding clientele,” said Stephen Roche, Hemscott’s CTO. “Speed is an absolute requirement.”

Enter XML accelerators
In the last few months, a new class of product, the XML accelerator, has started to emerge. Similar to the commonly used SSL accelerators, XML accelerators offload XML data-handling tasks from an application server.

The new products come from a variety of vendors and have a wide range of functions.

DataPower Technology’s XA35 XML Accelerator is an appliance that mainly handles XML transformation and conversion tasks.

Sarvega’s XPE 2000 Switch can accelerate XML data as well as intelligently route data based on information within a data stream.

Expand Networks’ Accelerator, which optimizes wide area networking links, improves the performance of Web services applications by handling XML-based SOAP (simple object access protocol) messaging data more efficiently.

To avoid server drain—a potential XML performance issue for Hemscott—Roche installed the DataPower XA35 XML accelerator. The XA35 converted the XML data to HTML format, thus offloading this task from the Web server. The dedicated XML acceleration hardware offered a significant improvement in the conversion time, said the tech leader. Conversions that used to take 20 seconds to perform now take only one second. But keep in mind that such a performance boost is not cheap—pricing for the XA35 starts at $55,000.