ACADEMIA

Oslo, Norway – Scali AS, a leader in high performance clustering solutions, announced today that they have completed the installation of a high performance Linux-based cluster at Storm Weather Center to run advanced parallel computing tasks. The cluster is built with 13 dual 2.8 GHz Dell 2650 PowerEdge nodes using high-speed Gigabit Ethernet interconnections with a maximum performance of 291 Gflops/second (single precision). Hardware components were supplied by Dell, while Scali provided software, system installation, application tuning services and ongoing support. “Meteorological calculations are extremely demanding and complex. We set strict requirements to the software and the professional competence of our partners” says Svenn Owe Haugland, Director of Research & Development at Storm. “Scali was chosen after a thorough evaluation where reliability, price, performance and professional services were key requirements. We are pleased with the attention and follow-up that Scali has provided and we look forward to working with Scali in the future.” “We see growing interest for high performance clusters from the weather industry where advanced systems are required to conduct accurate analyses with optimized performance for real-time calculation processing,” says Bjørn Skare, CEO at Scali. “Scali’s cluster experience combined with Dell’s hardware expertise allowed us to provide a complete solution to Storm that was up-and-running reliably in a minimum amount of time.” Storm is the most recent addition to Scali’s list of customers performing weather-forecast calculations on Linux-based, high performance computing clusters. Customers include The Norwegian Meteorological Institute (Oslo), The National Supercomputing Center of Sweden (Lindköping), The US Naval Air Systems Command Weapons Division (Point Mugu, California) and the University of Chile (Santiago). The Storm cluster runs on Red Hat Linux and is powered by the Scali Manage™ and Scali MPI Connect™ for Ethernet. A key requirement was the ability to guarantee uptime and reliability. All critical system components, including switches, interconnects, cabling and power sources were duplicated in order to achieve maximum failover capabilities.