August - 2003 issue > On The Cover

By Karthik Sundaram

IT CAN BE A BLADE. IT CAN BE A CHIP. IT CAN ALSO BE PURE code. Today, it is a box called the XPE 2000 sitting above the network and performing cop duty—capturing incoming messages, reading off the XML schema on it, and then sending it along forward, based on a set of security codes.

At $60,000 for a 60lb piece of hardware and software, the box, claim the three co-founders of Burr Ridge, IL-based Sarvega, saves millions for the enterprise in delivering wire-speed XML processing—a key activity in the application server environment. “XML trades performance for extensibility,” says Ted Schadler, principal analyst for software at Forrester Research. “The extensibility is huge because you can add mechanisms like security and encryption incrementally, but that means you have to parse the message to pull out the data. It’s a huge amount of overhead.”

In 1999, Sunil Gaitonde, John Chirapurath and Girish Juneja met at a TiE chapter meeting in Chicago. They found very complimentary backgrounds with each other—Gaitonde came from networking experience, Juneja had led technology teams at enterprises, and Chirapurath had marketed enterprise products. While the enthusiasm to embark on the entrepreneurial road was high, each was carefully evaluating the other, as the three explored possibilities for a business. “Distributed computing has been around for a while, and at the time we met it was taking on some unprecedent spotlight,” recalls Chirapurath, now the VP of marketing at Sarvega. Sufficiently convinced that there was a good challenge in open standard, the trio founded Sarvega, to find a solution for accelerated processing capabilities in the network environment. In a world based on XML Web services, XML processing will consume enormous amounts of CPU, memory, and network bandwidth. Especially in a transactional environment, the performance burden imposed by this new class of overhead can make applications unusable, says Larry Seltzer, an industry expert on the processor space.

Birth of the box
For over nine months, the trio wrote and re-wrote business plans and worked on the alpha version of their product. “As the network is mired in the multiple processing demand environment, solutions that could accelerate specific applications was seen to take importance,” says Gaitonde, a former CISCO networking man. As XML came into the limelight as the magic medicine for open standard in the network architecture and application convergence, there seemed like a good opportunity to develop a specific hardware for XML processing. The initial strategies to expand the network processing capabilities was by simply adding more servers— “throwing more UNIX and Windows boxes at the network,” laughs Chirapurath. This didn’t go very far and the market for specialized hardware came into existence. This, seemingly, is a natural evolution as many CPU-intensive tasks have been found to be executed better from a hardware than by a software.

In September 2001, Sarvega fetched its first round of funding. The post-9/11 investment environment wasn’t exactly rosy; Sarvega won on its strength of the technology and management capabilities. Much-quoted in print, Gaitonde sets out his pet process; he developed a three-prong evaluation tool: the laugh test, the market test and the money test. “If people laugh at my idea when I ask them if they would pay for it, it isn't going to work,” he says. The idea also has to be in a large enough market to have lots of sales and the VCs have to be willing to fund it. One has to be open to change as well. An experienced entrepreneur who sold his last company, Internet Junction, to Cisco, Gaitonde candidly admits that the only thing that remained from the original Sarvega business plan when it finally secured funding “is one half of its name,”—it used to be called Sarvega Wireless. So how did Sarvega attract new investors to the table in this market? “The fact that they had beta customers lined up to use their product was pivotal in our decision to invest,” says Bob Zieserl, managing director of KB Partners, which co-led the round with East Coast-based BessemerVentures. “We also talked to a lot of others and gavethem the Sarvega pitch and they said they would buy. That really validated the market potential forus,” he says, adding that such feedback is critical to evaluating an emerging market. “Our technology and IP could take any shape and form,” comments Juneja, VP of technology at Sarvega. “Sarvega’s patent-pending XML EventStream Operating System (XESOS™) technology is designed with two key objectives: performance and flexibility. Today it is a transparent hardware that sits in front of the network to stream the XML processing. Tomorrow we could deliver the same functionality and power from a chip, or even a blade that goes into the CPU.” The hard-won funding has been judiciously used in the product development and the startup was quick to find buyers. “Finding clients was not very difficult as many of our current customers had a big say in the XPE 2000’s evolution,” says Chirapurath.

If you have more query about magazine please send mail to

May 12	April 12	March 12	Feb 12