Northrop Grumman is a global defense and technology company specializing in information services, electronics, aerospace, and shipbuilding for government and commercial purposes. Northrop Grumman scientists and engineers place heavy computational demands on their computer systems. Simulations and computer-aided design applications require a lot of processing power, which Northrop Grumman met for years with multiprocessor workstations and small clusters of servers. Each project in Northrop Grumman received enough computing power to support its needs. Unfortunately, budgets did not always accommodate the needs of new projects, and if they did, setup was time consuming, and sometimes deals were lost in the delay. Maintaining many disjoint systems was also a major challenge for Northrop Grumman’s information system support staff. Bradley Furukawa, VP and CIO, had an idea for a better system. Rather than custom-designing, building, and supporting many computer systems for many projects, Furukawa wanted to build one large supercomputer cluster that could be shared by all projects. Furukawa was given permission to try his theory out on Northrop Grumman’s Space Technology unit.
Furukawa assembled a team of information systems specialists and scientists to create a massively parallel supercomputer from Linux-based blade servers. The cluster quickly grew to 979 processors supporting over 100 applications specially adapted for parallel processing. Furukawa saw his role in the development as more political than technical. "It was my job to make sure the funds were there, make sure [the project] stayed visible in front of the vice president and president, remove any administrative barriers ... and let the engineers and scientists do their thing."
Furukawa also had important diplomatic responsibilities. He worked to make sure that the designers of the system kept an “enterprise perspective” rather than focusing on their own project’s needs. A governance board was created to decide how supercomputer resources were to be shared across projects in the enterprise. "No matter how much computing power you have, you can always max it out. They set the priorities," Furukawa says. The board was responsible for creating the controls to allocate system resources and monitor their use.
Furukawa’s plan worked. The shared supercomputer cluster allows Northrop Grumman employees to complete work faster. Equally important, it allows new project proposals to get off the ground quickly. The salesforce now closes more new deals and keeps existing contracts funded. Customers appreciate the additional simulations and analysis provided by the increase in processing power.
Word of the success of the new supercomputer system reached top executives, who were so impressed that they funded additional growth for the system—which now has 1,800 processors running 400 custom designed scientific and engineering applications. Woffrking to keep the money flowing, Furukawa has opened up the supercomputer to other areas of Northrop Grumman over the company’s internal network. He wants to keep the system fully occupied to make a case for further expansion. The ultimate goal for the system is 3,000 CPUs.
Furukawa cites visibility as critical to the project’s success. The team was quick to get scientists working on the supercomputer as soon as it was ready. As scientists and their projects benefited from the system, word of the success spread quickly. "Keeping the project on schedule and within budget really added to the success and credibility," says Clayton Kau, vice president and general manager of Northrop Grumman’s space and defense products division.

Discussion Questions:
1. What technique was used at Northrop Grumman to assist scientists and engineers in being more productive?
2. What role did the CIO play in the development of Northrop Grumman’s new system?
1. Why was it smart to establish a governance board to manage the new system?
2. What do you think the CIO’s biggest challenges are in growing the new system? How is he meeting those challenges?