STORAGE

Virtualized Data Management Solution to Put Data on Most Cost-Effective Storage While Speeding Results for Most Complex Problems: Scientific and engineering problems are growing more complex, and that means faster computational power alone cannot always deliver results sooner. Facing this productivity stalemate, researchers at the University of Utah's Scientific Computing and Imaging (SCI) Institute discovered their problem had only one solution: a world-class storage and data management system from SGI. Expected to be delivered this spring, a new SGI InfiniteStorage virtualized data management solution will allow the SCI Institute to store and seamlessly manage more than 115 Terabytes of scientific data at the outset. Ordered in March with support from a National Science Foundation computing research infrastructure award, the new solution will help faculty, students and research staff strategically access information essential to their projects — a growing challenge in an environment where researchers in multiple disciplines work with the same source data, often at the same time. The institute expects the SGI solution will help deliver answers to even the most complex problems faster and more efficiently, while opening the door to promising new avenues of study. "Our goal was to implement an SGI solution to create a collaborative environment that gives our researchers on-demand accessibility throughout the entire process," said Dr. Steven Parker, an assistant professor in the School of Computing at the University of Utah and a long-time member of the SCI Institute, an internationally recognized leader in visualization, scientific computing, and image analysis. "For instance, a group within the SCI Institute might create a large simulation on one of our computing clusters, while another group — possibly weeks later — uses the same data set to run a variety of numerical or image analysis algorithms," Parker said. "Yet another group might use those results to fine-tune or test a new visualization method." Parker said the SCI Institute's existing storage facility renders this scenario impossible. "Our current resources don't allow us to maintain readily accessible large simulations and datasets, and that limits our research, and the impact of that research." A range of projects will rely on the new virtualized data management environment, which will anchor the high-performance computing (HPC) resources housed in the University's new John E. and Marva M. Warnock Engineering Building, which is opening this spring. "Our faculty and students are working on projects funded by both federal agencies and corporate sponsors, and they run the gamut of disciplines from biomedical imaging to visualizing nuclear fusion simulation data," said Parker. "Each of them will benefit significantly from this virtualized data storage environment, and often in very different ways. Some will be able to develop new algorithms and systems that today may not seem feasible, while others can more easily collaborate with researchers outside the university." For SCI Institute researchers, data management and access are simplified with SGI InfiniteStorage Data Migration Facility (DMF), which allows datasets to be transparently migrated between high-speed short-term storage and lower-speed long-term storage, and SGI InfiniteStorage Shared Filesystem CXFS for sharing data between multiple operating systems within the SCI Institute compute facility. Further, SGI InfiniteStorage NAS Manager software will allow SCI Institute administrators to perform diagnostics and keep close track of performance and disk utilization via a Web-based graphical user interface. Built on a SGI InfiniteStorage NAS 4550 solution, the SCI Institute environment is designed to seamlessly move aging or low-priority data off of the 38.4TB of primary high-performance Fibre Channel storage to a more economical 40TB of secondary Serial-ATA on an InfiniteStorage 4500 RAID array. Eventually, the lowest priority data moves to a 40TB StorageTek SL500 tape library. As the SCI Institute's needs warrant, the InfiniteStorage NAS platform can scale its support for NFS and CIFS users, and for ever-larger data storage capacity. James River Technical, Inc., SGI's designated value-added reseller for higher education, participated in the sale. "The scalable and reliable storage hardware is absolutely critical, but SGI's unique software component allows truly virtualized storage and data management," said Parker. "CXFS and DMF were major drivers for our selection of SGI. These solutions will ensure that all data in the system will remain essentially ‘live' and visible to users." "The SCI Institute has again proven itself to be a forward-looking research facility, not only for its groundbreaking areas of study, but also for its strategic vision in establishing a virtualized storage management system that is truly state of the art," said Raj Das, vice president, storage, SGI. "This institution is wisely leveraging SGI storage and data management resources to remain ahead of the curve on rapidly growing datasets and increasingly complex problems and algorithms. We look forward to working with the SCI Institute in the future as its needs evolve."