ACADEMIA

DataDirect's S2A Storage Systems Enables 6 of the Top 10 Fastest Clusters in the World: DataDirect Networks, the leader in high performance, high capacity storage solutions, has extended its leadership role with the recent release of the TOP500 Supercomputer Sites List. The company continues to be the largest supplier of high performance storage to the world's top 100 cluster sites, powering 6 of the top 10, 24 of the top 50, and 53 of the top 100 fastest machines in the world. DataDirect Networks' explosive growth and acceptance in the HPC (High Performance Computing) community is driven by its unrivaled Silicon Storage Architecture (S2A) technology. S2A cluster storage solutions deliver in excess of 100 GB/s aggregate bandwidth to and from the disk array with petabytes of capacity, while maintaining open-systems infrastructure support to power computer clusters from IBM, Dell, Cray, SGI, Bull and others. DataDirect's S2A Storage Appliances enabled 32% of the 4,947 teraflops processed on the TOP500 list, including powering the world's fastest supercomputing cluster, BlueGene/L, a National Nuclear Security Administration system at Lawrence Livermore National Laboratory. DataDirect also has major installations at Oak Ridge National Laboratory, Sandia National Laboratories, the New York Center for Computational Science, CEA, NASA, Atomic Weapons Establishment, Caltech, Cambridge University, NCSA, Swiss Scientific Computing Center (CSCS), US Army Research Laboratory (ARL), UCSD, CINECA, Indiana University and the University of Texas among others. "Ever increasing demands on Livermore's scientific computing capabilities required a storage solution that offered a significant jump in I/O scalability, reliability, performance and the ability to meet deadlines for delivery and integration," said Mark Seager, Assistant Department Head for Advanced Technology at Lawrence Livermore National Laboratory. "We needed a cost-effective storage solution that could meet the production demands of scientific computing at the BlueGene/L scale. Predictable performance under intense I/O workloads coupled with reliable hardware and software for our Lustre-based simulation environment enabled us to quickly meet the computational demands of programs critical to national security." "There is a reason that DataDirect has been so successful in the top cluster sites powering the fastest machines in the world," added Tony Asaro, senior analyst, Enterprise Strategy Group. "Their Silicon Storage Architecture is purpose-built for high performance. We are seeing the need for this kind of speed becoming pervasive beyond the HPC world and finding its way into mainstream data centers. DataDirect has implemented innovation at a system level and their deep understanding of storage architectures will drive continued success as the demands placed upon customer environments continue to increase." DataDirect has also recently released their exclusive SATAssure technology for the S2A to address the growing gap between higher performance and capacity needs of clusters vs. the need to save money. SATAssure allows HPC users to benefit from the capacity and cost advantages of deploying economical SATA drives in large, scalable storage pools without the risk of data corruption or loss, while still achieving throughput targets in both reads and writes. SATAssure also reduces the need to perform lengthy drive rebuilds due to SATA disk drive errors, lowering drive replacement rates and the risk of data loss. Last, DataDirect's S2A Appliance eliminates the I/O bottleneck impasse that has plagued the HPC community. The S2A's unique parallel access design creates a common, shared storage pool that accelerates cluster operations by allowing multiple CPUs to access the same information at the same time, allowing more tasks to run in parallel and in less time. The S2A collapses the storage network into fewer hardware and software components through the integration of switching, virtualization, RAID, intelligent caching, and disk-level optimizations into a single device, simplifying and cost reducing the storage network while boosting its performance and making it easier to administer. The S2A delivers real-time sustained performance even in the face of challenging environments with multiple simultaneous host requests and mixed reads and writes to the array, without performance penalty even in the face of disk errors or failures, helping cluster simulations complete faster, deliver faster results analysis, and improve the ROI of the entire cluster by keeping the CPU nodes busy solving problems instead of waiting for the storage system to respond.