SCIENCE

ANNAPOLIS, MD -- Scyld Computing Corporation today released the Scyld Beowulf Professional Edition, the latest version of its next generation cluster operating system software. Professional Scyld Beowulf greatly simplifies cluster setup, integration and administration, while providing seamless scalability. Coupled with documentation and support from the original Beowulf development team, Professional Scyld Beowulf provides the first true clustering solution that can be installed and run directly out of the box. Beowulf cluster systems connect a series of computers together, using a modified version of Linux, to form a parallel processing supercomputer. The Scyld Beowulf Operating System improves upon traditional Beowulf clusters as all operations performed on the linked cluster nodes are initiated and administered through a single master node. Its Single System Image (SSI) design makes the cluster act and feel like a single computer. This drastically simplifies maintenance and eliminates common pitfalls of clustering such as version skew and runaway jobs. Scyld Beowulf only needs to be installed on one master node and will run clusters of hundreds of compute nodes. Professional Scyld Beowulf is a complete software system. No other software is required to create the cluster. For industries such as energy, biotechnology and finance, there is growing commercial demand for High Performance Computing (HPC) clusters. Companies are looking for more cost-effective ways to analyze enormous amounts of data. One example of this demand is in the petroleum industry, where 3-D seismic modeling is used to locate oil fields. Because large amounts of data must be processed to create an image of a complex geology, seismic modeling is a very compute-intensive process. Such imaging turns raw, unprocessed data into a coherent image that accurately depicts the situation below the earth's surface, saving millions in reduced exploration and drilling costs and improved production. Scyld Beowulf has distinct price advantages over traditional Symmetric Multiprocessing (SMP) or vector supercomputers. Using commodity servers and hardware components, a Scyld Beowulf based system can provide the same amount of computing power at 10%-50% of the cost. Continued savings are realized by the reduced costs of supporting a broad-based operating system as compared to the expensive maintenance contracts required for proprietary operating systems of traditional SMP or vector supercomputers. Companies can save even more by leveraging their existing hardware and installing Scyld Beowulf, allowing them to set up a cluster system for a fraction of the cost of traditional solutions. In addition, Scyld Beowulf's simplified master node installation and administration eliminates the administrative costs and potential risks inherent in other HPC clusters that require nodes to be administered separately. Donald Becker, Scyld's founder and Chief Technology Officer, and other Scyld developers were the original architects of Beowulf computing while working at NASA as research scientists. That same team has improved and adapted the technology for the commercial market, using the same exacting standards for software engineering, development, quality control, test methodologies and support that NASA uses to ensure successful missions. The result is Professional Scyld Beowulf. "Our mission at Scyld is to create software that will bring cost effective, easily managed, high-performance computing to the commercial marketplace", said Becker. "The new features in this release add significantly to our existing standard upon which high performance cluster applications have been developed. This new release will further stimulate deployment of turn-key commercial applications". Scyld has formed and is in the process of forming partnerships and alliances with many clustering industry leaders including, Compaq, Arrow/Wyle, API, Penguin Computing, GIGABYTE Server Group, PSSC Labs, Aspen Systems, Atipa Technologies, RLX Technologies, CFI, Racksaver, eLinux, Cendio Systems, Western Scientific, VA Linux, and Myricom, amongst others. Scyld works closely with its partner's hardware systems to certify seamless operation of hardware platforms such as Compaq's Proliant DL380 and DL360 servers, Compaq's DS10 series AlphaServers, API's CS20 and UP2000+, Penguin Computing's servers, RLX Technologies' System 324 Blade technology, and Intel's ServerBoards. In addition, Scyld has partners and alliances with traditional parallel application and tool providers such as, MPI Software Technology Inc., Veridian PBS-Pro, Wolfram Mathematica, NAG, Absoft, Lahey, Backbone Networks, NAMD, CHARMM, and TurboGenomics. Scyld has a formal channel program to authorize and train value added resellers to provide off-the-shelf, fully integrated and supported turn-key cluster systems. New enhancements on the latest Professional Version include, full Alpha support including simplified installation tools, full Myrinet and Gigabit Ethernet support, the Scyld Beowulf Batch Queue system (BBQ), automatic node addition, web based administration and job monitoring, advanced hardware health and status monitoring, Parallel Virtual File System (PVFS), NFSv3, and ROMIO file systems, library, and much more. For more information visit the Scyld website at www.scyld.com