RAINFINITY: From Outer Space to Interspace
By Michael Rogers

Demonstrating RAIN technology in Shuki Bruck’s lab, graduate student Massimo Franceschetti unplugs one of five interconnected servers that operates an on-screen movie. RAIN software is able to instantaneously reroute connections among the servers to keep the movie playing without interruption.

Rainfinity traces its origins to 1994, when Shuki Bruck, professor of computation and neural systems and electrical engineering at Caltech, and Leon Alkalai, director of JPL’s Center for Integrated Space Microsystems, came up with an idea to improve NASA’s computer systems, both on the ground and in space.

“In the past, NASA would customize every computer component—both hardware and software—for each job, which is extremely expensive,” says Bruck, an expert in parallel and distributed computing and fault-tolerant computing. “It made sense to us to propose a project to use commercial, off-the-shelf components.”

Bruck and Alkalai went to Washington, pitched the idea to officials with the Defense Advanced Research Projects Agency (DARPA), and by 1995 got funding from NASA and DARPA to develop an alternative to the costly, special-purpose computer systems used in space missions. Bruck gathered a team of five graduate students, and by 1997 they had built a prototype of a system called the Redundant Array of Independent Nodes, or RAIN.

One of the requirements for building an off-the-shelf computer system for NASA is that it must be extremely reliable, since the consequences of a malfunction can be catastrophic in space. In the RAIN system, there are multiple computer nodes and connections that can perform the same function.Rather than design expensive fault-tolerant chips, Bruck and his team designed a system so that if one chip failed, another one would take over. In this system, a component that stores data across distributed processors retrieves it even if some of the processors fail. If part of a system shuts down, the recovery by another part of the system is instantaneous and does not disrupt any operations.

The RAIN prototype was finished at a time when more people had begun going online, to shop or check out companies and products, and businesses began to recognize the Internet as a tool that was vital to their future. Bruck realized that companies that relied on the Web for business were subject to dire consequences if their systems crashed, since competition was now just a click away. And Internet sites crashed often because the computer systems handling Internet connections were not built to manage the Internet onslaught. Bruck figured that RAIN could be applied to the Internet and he began investigating the possibility of starting a company to commercialize RAIN.

In the spring of 1998, Bruck approached the five graduate students who had worked on RAIN and asked them if they wanted to start a company with him. Two opted to pursue academic careers, but three signed on—Vincent Bohossian, PhD ’98, Charles Fan, PhD ’01, and Paul LeMahieu, MS ’96. In addition, Phil Love, PhD ’99, then a graduate student in applied mathematics, was recruited for the team.

Bruck, together with his business partners, then found several investors to provide a total of $2 million to launch the company, which they named Rain-finity. With the money, they set up a research office in the Old Town section of Pasadena in September 1998. They then began creating a software product so that companies would no longer have to rely on single Internet gateways to their Web sites. A com-pany’s Web site could be accessed faster, and many more people could get into the site at the same time. There would be multiple pathways to route traffic, and the system would also serve as a so--called firewall against outside security breaches and viruses. The system, called Rain-wall, was completed in 1999. It sells for $5,000 to $20,000, depending on the number of processors supported.

“RAIN technology is reliable software technology with high availability,” says Bruck. “It’s like a Borg from Star Trek. If you kill a machine it keeps functioning. Others take over. And users don’t see the effect of that. We do have competition. But our technology is one generation ahead.”

While starting the company, Bruck also negotiated with Caltech, which owned the rights to the RAIN tech-nology since it had been created at Caltech. The Institute got equity in the company in exchange for the rights.

By the summer of 1999, Rainfinity had its first paying customer: the Chicago Board of Trade. Bruck then raised $15 million through venture capitalists, and hired a chief executive officer: Olivier Helleboid, who formerly had run the largest software division at Hewlett-Packard. Rainfinity opened its headquarters in San Jose in the summer of 1999 so that it could be close to a strong employee base and to potential corporate partners. It kept its research arm in Pasadena, since many of its employees continued to have connections to Caltech.

Once Helleboid was on board, Bruck—who serves as chairman of the company—relinquished his involvement in day-to-day operations to focus more on his own research at Caltech. “My role now with Rainfinity is either as cheerleader or pain in the rear,” jokes Bruck. The Caltech graduate students who helped start the company with Bruck now have a variety of engineering and managerial roles with Rainfinity.

For Charles Fan, director of technology at Rainfinity, who came to the United States in 1989 from China and started his career in business as a paperboy while in high school in Indiana, helping start Rainfinity has been exhausting but thrilling. While developing Rainfinity’s products and refining the technology, he has also been responsible for recruiting and hiring technical employees—many of whom have come from Caltech—and managing the engineering side of the business.

Once customers started buying the product, Fan also found himself flying around the country and the world at a moment’s notice to install systems or assist in troubleshooting. Adding to his hectic schedule, he got married to Fang Wang, PhD ’98, and they had a baby daughter in September 2000.

“I’m extremely proud of the technologies that we’ve created at Caltech and at Rainfinity,” said Fan, who took a leave of absence from Caltech to work at Rain-finity before receiving his PhD in 2001. “Ten years from now, I hope people will see that we created something useful and solved people’s problems.”

Rainfinity, which had less than 20 employees in 1999, has now opened offices in England, Germany, France, and Brazil, and has more than 200 customers, including Dresdner Bank in Germany, Boise Cascade, and Qantas Airways.

The financial implications of this expansion only partly motivate Fan. “Making a lot of money is one of the excitements of the start-up experience, but it is not the only motivation,” he says. “It would be a big mistake if we were driven by money in terms of company strategy. The company must remain focused on customer needs.”

Indeed, given the fluid nature of the Internet, there should be a continued need for Rainfinity’s products. Fan predicts that new Internet issues will arise involving security, reliability, speed of service, and versatility.

“As the Internet evolves, there will be new needs and solutions that address these needs,” Fan says. “There will be holes and bugs in those solutions that create new problems. Then there will be new solutions to solve these new problems. It’s a continuing cycle and a continuing process in how the Internet grows. When I came to Caltech, I was young and wanted to learn as much as I could. This is more than I hoped for, and I’m having a blast.”

Article Links:

Adventures in Entrepreneurism

Ortel in Three Acts

Clinical Micro Sensors: Star Trek Meets the Human Genome