Lawrence Livermore National Labs Selects DataDirect

DataDirect Networks, Inc. announced that Lawrence Livermore National Labs (LLNL) has selected the company’s S2A9900 data infrastructure platform to provide scalable I/O for its next-generation computational simulation environment. The storage will be used by a number of LLNL supercomputers, including its newest system, Dawn. In addition, DDN storage will be utilized in a new project, Hyperion, to evaluate scalable storage architectures capable of more than one terabyte per second.

Lawrence Livermore National Labs

Dawn is the first of two scheduled system additions that will increase the lab’s computational resources. Initially expected to achieve 500 teraFLOPs per second, the solution utilizes 16 DataDirect Networks’ S2A9900 high performance storage platforms. The S2A platforms are the foundation for a site-wide Lustre file system environment, which provides a dynamically scalable storage infrastructure that delivers mission-critical uptime, maximum performance and scalability across tens of thousands of drives.

"The high performance scientific computing we need to fulfill our national security missions requires industry-leading storage density, optimized performance for scalable compute environments, real-time RAID 6 for maximum data protection, reliability and the quality-of-service needed for massive data environments," said Dr. Mark Seager, Assistant Department Head for Advanced Technology at LLNL. "Previous experience has shown us that high- quality, dedicated support in helping us achieve our program objectives for performance, reliability and usability is an integral part of this partnership — a partnership enabling the lab to succeed in its mission, under tight timelines and extreme pressure."

As further validation of the lab’s scale-out storage strategy, LLNL recently received a ‘Best Practices in Storage’ Award from Computerworld‘s Storage Networking World for successfully deploying one of the world’s fastest supercomputers, BlueGene/L, based upon DataDirect Networks’ S2A storage and the Lustre file system.

"Lawrence Livermore is a perfect example of why our Silicon Storage Architecture is being utilized by half of the Top100 supercomputers in the world and why that number continues to increase every year," said Alex Bouzari, CEO and Co-Founder of DataDirect Networks. "We’ve always believed that our technology provides our customers with a distinct performance and capacity advantage, as it was built from the ground up to handle the world’s most demanding applications. The fact that Lawrence Livermore has once again selected DataDirect Networks further demonstrates our leadership in this industry. As the only HPC solution provider to consistently deliver storage platforms with extreme levels of performance and scalable capacity required to address high-speed scientific simulation, computation, data analysis and visualization, we look forward to helping LLNL fulfill its mission and to continue pushing the boundaries of performance and innovation."

LLNL and DataDirect Networks will extend the success of their award- winning partnership in a new project, called Hyperion, which brings together 11 leaders in the computing industry to jointly deploy a large-scale Linux cluster testbed. DataDirect Networks and LLNL will focus on using Hyperion as a Petascale I/O testbed to evaluate scalable storage architectures capable of more than one terabyte per second. The first half of Hyperion is now on-line and being used by the collaboration. When fully deployed in March, 2009, the Hyperion cluster will have at least 1,152 nodes with 9,216 cores; with about 100 peak teraFLOPs per second; over nine terabytes of memory; an InfiniBand DDR interconnect and access to over 13 gigabytes per second of storage bandwidth based on DataDirect Networks’ S2A high performance storage platforms.

With over 160 petabytes installed worldwide, DataDirect Networks’ S2A solutions deliver more aggregate bandwidth to the Top500 Supercomputers than any other IT vendor, including six of the world’s Top10, 28 of the Top 50 and 48 of the Top100 fastest supercomputers.