CSC-Supported Supercomputer Assists NASA's Return to Flight

Client: NASA Advanced Supercomputing (NAS) Division, NASA Ames Research Center

Challenge: To work closely with prime contractor Advanced Management Technology, Inc. (AMTI) to build and integrate NASA's Columbia supercomputer in 120 days.

Solution: With expertise in fields ranging from management to high-end computing, the joint AMTI-CSC teams, working with NASA and industry partners, successfully completed the complex tasks required to integrate Columbia within the aggressive timeframe stipulated by NASA.

Results: Columbia received a top LINPACK rating of 51.9 teraflops (51.9 trillion floating point operations per second), making it the world's fastest production supercomputer. Also, the supercomputer is supporting more than 650 NASA users across an array of vital projects including Return to Flight, aerospace vehicle design, and climate research and ocean modeling and the CSC team continues to provide critical systems, network, applications, and operations support for Columbia and its users.

For more information about the Columbia Supercomputer integration, contact CSC's Chris Buchanan, 650.604.4308, cbuchana@nas.nasa.gov.

For additional defense-related information, please see CSC's Government section.

Contact us and let our experience help you produce results.

When the space shuttle Discovery touched down safely on August 9, 2005, capping a successful 14-day mission that marked NASA's return to manned space flight, it was due in part to a CSC-supported supercomputer that helped make the shuttle safe for landing.

Before the shuttle could land at Edwards Air Force Base in California, NASA scientists and astronauts had to find and repair possible damage to the shuttle sustained during takeoff. They used the CSC-supported Columbia supercomputer housed at the NASA Advanced Supercomputing Division at NASA Ames Research Center in Moffett Field, Calif. Named to honor the crew of the space shuttle Columbia lost in February 2003, the supercomputer played an integral role in the Discovery mission, providing continuous technical analysis from launch to landing. Capable of performing 51.9 trillion floating point operations per second, Columbia is among the three fastest supercomputers and is the largest Linux-based shared memory system.

Working around the clock

Prior to the launch, scientists used the Columbia supercomputer to conduct detailed analyses of the kind of damage falling debris from the fuel tank could cause to the shuttle. This included developing an automated plotting tool for predicting debris paths and determining how hard and damaging the debris impacts might be. During the mission, scientists at Ames used the supercomputer to determine that errant pieces of gap filler cloth between Discovery's heat-shield tiles could cause the shuttle to overheat upon re-entry to the earth's atmosphere. This assessment was key to NASA's decision to undertake an unprecedented space walk to make emergency repairs to remove the cloth from the underbelly of the spacecraft.

Teaming up with NASA and the program's prime contractor, Advanced Management Technology, Inc., CSC professionals helped design, build and integrate the Columbia supercomputer in the summer and fall of 2004. Working virtually around the clock, the team built and integrated the supercomputer to meet NASA's aggressive schedule of completing the implementation in just four months. CSC continues to provide day-to-day operational support for the supercomputer, including applications, network and help desk services.

Supercomputer leverages teamwork

The Columbia Supercomputer increases NASA's computing capacity ten-fold. The 10,240-processor system provides an integrated computing, visualization, and data storage environment, enabling America's premier space agency to achieve breakthroughs in areas such as advancing aerospace design and shuttle modeling to prepare the Space Shuttle for safe return to space, space exploration, and high-resolution climate and ocean modeling.

Says Chris Buchanan, CSC's site manager and network technical lead, "It was an extraordinary effort - in effect, doing the impossible. To coordinate an effort on this scale demanded dedicated expertise and specialized skills. Fortunately, CSC's high-end computing and networking groups brought exactly the skills and experience that AMTI and NASA needed."

"More than anything else, though," Buchanan continues, "this is a story of partnership, of teamwork. Every single team included AMTI and CSC employees. We worked shoulder-to-shoulder through the entire build and integration of Columbia."

The challenges

Other than the time crunch, one of the biggest challenges the teams faced was the fact that Columbia had to be built and integrated into a production environment without shutting the entire facility down. Explains Buchanan, "Through these 120 days, even as we were receiving and integrating the 20 Columbia nodes, we had to keep the facility up and running and our user base happy. This meant that as we removed the other supercomputers to make room for Columbia, we still had to ensure NASA users could run their jobs with minimal interruption. That demanded a level of expertise and time-to-solution which compounded all our other challenges. It wasn't just a matter of installing a piece of hardware, plugging it in, and switching it on. The architecture for eight of the nodes making up this system was first-of-its-kind. There's a lot that had to happen to make this project a success, and we're proud CSC - working closely with AMTI, NASA, SGI, Intel, and other NASA partners - was an integral part in making it happen."

Today, the Columbia supercomputer is supporting more than 650 NASA users across an array of vital projects, including "Return to Flight" efforts; NASA Engineering and Safety Center activities; aerospace vehicle design; climate research and hurricane prediction; and space exploration. The CSC team continues to provide day-to-day operations support for the supercomputer, including a 24/7 help desk; 24/7 systems, network, applications, and production support; and continual enhancements to systems and network integration and design.

The CSC team is also helping build a high-speed, 10-gigabit wide area network infrastructure to link Columbia to the other NASA centers.