IBM Brings Supercomputing To Hollywood

Amsterdam (The International Broadcasting Convention), Media producers, digital editors and animators worldwide can now collaborate, edit and render animation and films by harnessing the power of supercomputing technology using IBM's General Parallel File System (GPFS). Supercomputing is the ability to compile millions of computations per second, and often features hundreds of computers working together for a single application. GPFS is a high-performance file system that helps power supercomputers via its massive storage capabilities. A new, core component of IBM's digital media technology architecture -- GPFS brings a whole new way of editing to the media and entertainment industry. With GPFS, digital content creators can quickly, easily, and simultaneously access, exchange, and edit rich media files. "GPFS allows multiple parties to get at the same content, at the same time," said Jurij Paraszczak, chief technology officer of IBM's Digital Media Group. "In our dealings with broadcasters and other media and entertainment companies, not only do people want to render animations and films at higher speeds, but they also want multiple people to be able to view and edit - GPFS does just that, even at high definition." GPFS is a huge leap in the speed of storage access and retrieval. A highly scalable system, GPFS is designed to manage terabytes of large files efficiently from a huge, shared pool of data storage. GPFS is available on IBM pSeries servers and runs on Linux Clusters and AIX. GPFS is a parallel cluster file system, based on a shared-disk model, which can handle the largest data needs. At customer sites, it currently delivers fast reliable access to up to 500 terabytes of data. (GPFS' architectural limit exceeds 2 petabytes, which is 2,000 terabytes or 2 million gigabytes). Because GPFS stores one single file copy, making it accessible by the entire system, using GPFS leads to several benefits and uses: • A master copy only needs to be created once, eliminating duplication costs • Multiple editors and viewers can access the same file instantaneously • Realtime editing of High-Definition compressed material can be handled • Large amounts of digital media can be stored and accessed at high speed • Less storage capacity is required, lowering hardware and software costs • Uses for GPFS include: content management, post production, collaborative editing, realtime ingest and viewing, and high performance computing. GPFS establishes a base storage framework for future open, industry standards as they evolve. GPFS media applications for Hollywood include "feed and edit/ingest" and infrastructure integration. The feed-and-edit/ingest capability delivers an unsurpassed speed advantage to critical news, sports and other digital-video editing applications. (GPFS has run at 9 GigaBytes/sec on LLNL's ACSI White system). This function enables broadcast video feeds to be simultaneously ingested into the shared storage pool, where different editors can simultaneously begin to edit it even before the feed download is completed. Conceived at IBM's Almaden Research Center in San Jose, Calif., Laboratory, GPFS is derived from the high-performance, "high-throughput" Tiger Shark file system developed for video archiving. Companies that are using it include Cable News Network (CNN), as part of its digital video archive and Lawrence Livermore National Laboratory (LLNL), which uses GPFS on one of the fastest computers in the world: the ASCI White Supercomputer. Now, its capabilities have been broadened to excel at handling large-scale multi-media and non-multimedia applications, such as scientific computing, data mining, digital library and scalable network files. IBM is now transferring these unique experiences to benefit GPFS' heritage -- scalable, multimedia video servers with GPFS for Media.