The origins of Nebula One go back to Kemp’s days at NASA, which he joined in 2006 as director of strategic business development. In 2007, he became a chief information officer, making him, at 29, the youngest senior executive in the U.S. government. In 2010, he became NASA’s chief technology officer. Kemp spent much his time at NASA developing more efficient data centers for the agency’s various computing efforts. He and a team of engineers built the early parts of what is now known as OpenStack, software that makes it possible to control an entire data center as one computer.
Palo Alto Research Center, or PARC, has three Nebula Ones, which it uses for research projects such as an effort to improve parking in big cities. Researchers at PARC have been analyzing huge amounts of data to create models that show when workers, delivery vehicles, and shoppers tend to use certain parking spots. The idea is to create parking spots with modifiable, electronic signs that can turn, say, loading zones into regular parking spots over the course of a day. “You need to spin up a large simulation, get the results, present them, and then sit back for a while,” says Roger Hoover, an engineer at PARC. With the Nebula One, an engineer can shut down one simulation and then start up a new project in a few seconds. Similar things are possible on Amazon.com’s EC2 service or Microsoft’s Azure, but those services can be expensive and slow when you must send large amounts of data back and forth via the Internet. “What we do just will not work on Amazon,” says Surendra Reddy, who leads the cloud technology work at PARC.