Beowulf Cluster Computing Inducted into Space Technology Hall of Fame
Overview
In April 2022, Beowulf Cluster Computing was inducted into the Space Technology Hall of Fame during the 37th Space Symposium hosted by Space Foundation in Colorado Springs, Colorado. This international award recognizes the impact of an idea first conceived at NASA Goddard Space Flight Center in 1993.
Project Details
The Beowulf approach for constructing powerful computing clusters uses only commodity parts: personal computer boards, Ethernet interconnects, and open-source Linux. This idea was born in NASA Goddard’s Space Data and Computing Division, which provided supercomputer access to NASA scientists for analyzing satellite data and running their computer models. Providing top-of-the-line Cray supercomputer cycles to all of the NASA scientists who needed them was too costly. Beowulf clusters were 10 times cheaper.
Under the Federal High-Performance Computing and Communications (HPCC) Program, most of NASA’s Earth and Space Sciences Project effort was focused on gaining access to the biggest scalable parallel computers in the U.S. to put Grand Challenge scientists on a path to achieving teraflops computing. A less visible project milestone was to develop a workstation made of parallel computers that could sustain one gigaflops (billion floating-point operations per second) at a cost of $50,000. This milestone inspired the ideas that became Beowulf.
The Beowulf concept using all commodity hardware and open-source software came from then-NASA research scientist Thomas Sterling. Using commodity hardware meant that each system would have the lowest possible cost and allow multiple Beowulf systems, which in turn would enable many software developers. Teaming with Sterling on Beowulf development was Don Becker, whom Sterling recruited to NASA for the project; a small staff, often students of existing university collaborators; and NASA Goddard’s in-house team of computational scientists.
Results and Impact
The NASA Beowulf team’s ”How to Build a Beowulf” tutorials and the 1997 Red Hat release of multi-platform CDs of Beowulf Linux brought commodity tools for parallel computing to the mainstream for the first time, and the team went on to publish a popular book on this topic in 1999. NASA’s investment in Beowulf was around four people per year for four years. Following that, Beowulf took on a life of its own, and the NASA team became a contributor to the project, rather than its coordinator.
Why HPC Matters
After having proved successful at NASA, Beowulf Cluster Computing rapidly spread into supercomputing facilities globally. This inaugural method of building high-performance computing clusters has continued to be used for the past 20 years because it delivers superior price/performance. Currently, most of the world's TOP500 computer systems use the Beowulf approach.
What's Next
Space Foundation's recognition of Beowulf's lasting contribution to high-end computing, along with NASA Goddard's role in its initiation, has cast a public spotlight on the creative work of a group of technical collaborators who took Beowulf from concept in 1993 to an easily and widely replicable kit form by 1997. This new and welcome interest has re-energized this author to organize memories and records from that timeframe to construct a narrative that better brings into focus the breadth of work done during those four years and that identifies and honors the individual contributors. This narrative of innovation in computing technology will include its origins in the earliest days of space flight and will likely take the form of papers and book chapters.
Presentation
More Information
- The Roots of Beowulf
- Beowulf Clusters Make Supercomputing Accessible
- Satellites, the Space Race, and Supercomputing: How NASA Goddard's Beowulf Cluster Computer Became an Award-Winning Space Technology
Researchers
- James Fischer, NASA Goddard Space Flight Center Please enable Javascript to see email address.