INDUSTRY

HP today announced new product and partner updates to its high-performance computing Unified Cluster Portfolio that offer the maximum performance per dollar for Linux clusters. Introduced at the International Supercomputer Conference, updates to the Unified Cluster Portfolio were highlighted by HP StorageWorks Scalable File Share (HP SFS) 2.0, a Linux cluster file server that triples performance and doubles capacity at a lower entry price. Enhancements to HP XC System Software and HP Cluster Platform systems were also included. The HP Unified Cluster Portfolio is a modular package of hardware, software and services that helps customers better manage, store, access and visualize the large volumes of data created by the computational analysis used in such highly complex areas as human genomics and the development of nano-scale electronic devices. HP leads the high-performance computing (HPC) market in revenue with a 34 percent market share, according to first quarter 2005 figures released by IDC. This represents a more than five percentage point lead over IBM. "The HP Unified Cluster Portfolio provides us with an integrated, open source-based supercomputing utility to support distributed, high-performance computing, data management and visualization," said Hugh Couchman, scientific director, SHARCNET, an Ontario-based HPC institute that spans 11 leading academic institutions and exists to conduct leading-edge research in many areas of science and engineering. "Such technology is critical to Canada's competitiveness as a world-class research community and our ability to accelerate the production of results that are of benefit to our economy, health, environment, scientific knowledge and culture." Through participation with partners and the open source community, the Unified Cluster Portfolio is the first solution in the industry to provide a common implementation across multiple operating systems (HP-UX, Microsoft Windows(R) and Linux) and HP Integrity and HP ProLiant servers based on industry-standard processors (Intel(R) Xeon(TM) and Itanium(R) and AMD Opteron(TM)). "HP's Cluster Platforms offer customers greater simplicity, flexibility and increased performance at a lower cost, and with the latest version of HP SFS, HP is continuing to deliver on its StorageWorks Grid strategy," said Winston Prather, senior vice president and general manager, High Performance Computing Division, HP. "As the leading HPC vendor, HP is using industry-standard technology and open source to deliver HPC solutions at price points that are significantly less expensive than proprietary offerings." HP SFS 2.0: Maximum performance per dollar for Linux clusters HP StorageWorks Scalable File Share 2.0 provides scalable performance at half the entry-level price of the first version and one-fifth the price compared to scalable network file systems (NFSs). It also offers more than 35 gigabytes per second bandwidth and 512 terabyte capacities -- triple the bandwidth and double the capacity per file system of the previous version and 10 times greater than NFS-based scalable systems. Now with support for Voltaire's InfiniBand, HP SFS 2.0 provides the widest range of high-speed interconnect support, is easily installed in a few hours, and has expanded monitoring and management features. With HP SFS 2.0, HP is advancing the StorageWorks Grid -- a standards-based architecture that allows storage services, such as HP SFS or email archive retrieval, to be delivered across a massively scalable, centrally managed system. StorageWorks Grid divides storage, indexing, search and retrieval tasks across a distinct set of computing nodes or storage "smart cells" that cooperate to form a single shared file system. Using HPC-oriented HP StorageWorks SFS20 disk arrays, HP SFS 2.0 offers the best price/performance and scalability for Linux-cluster storage, providing high-performance computing users with faster, more accurate analysis with higher cluster throughput and fewer input/output bottlenecks. "HP SFS delivers the high-performance shared storage that we require to maximize the throughput in our HP supercomputing cluster," said Dr. Wilfried Juling, professor at the Faculty of Computer Science and director of the Scientific Supercomputing Center Karlsruhe in Germany. "We depend on this supercomputing power to accelerate the industrial and theoretical research we do for the universities of the state of Baden-Wurttenberg and our industrial partners Porsche AG and T-Systems, a leading German provider of computing services. We plan to grow this cluster to 11 teraflops in 2006. The capacity and bandwidth scalability that HP SFS 2.0 provides will allow us to match the planned increases in our computational cluster performance." Expanded HP Cluster Platforms The foundation of the updated Unified Cluster Portfolio consists of three standards-based, factory pre-assembled and tested HP Cluster Platforms. The expansion of the Cluster Platforms will support new Dual-Core AMD Opteron processor-based servers in the HP Cluster Platform 4000, along with the recently introduced HP ProLiant DL145 G2 servers. The HP Cluster Platform 3000 is also now available with HP ProLiant DL140 G2 servers. In addition, the family of Cluster Platforms is expected to support HP BladeSystem servers, starting in August with the HP Cluster Platform 4000BL that features HP ProLiant BL35P systems. HP XC System Software gains momentum with customers HP XC System Software is a unique combination of open source and commercial products for best-in-class cluster management. It includes a full Linux operating system, cluster management capabilities for installation and ongoing system administration, robust resource management and scheduling, and integrated HP-MPI (message passage interface) within a single, supported environment. This month, HP is releasing XC Version 2.1, which extends XC support to the latest HP Cluster Platforms, as well as continued ease-of-use and quality enhancements. Leading HPC partners broaden value of Unified Cluster Portfolio The availability of reference cluster solutions with a defined application interface and design simplifies development and support of applications for cluster deployments. More than 30 independent software vendors have committed support for the Unified Cluster Portfolio, including most of the leading vendors of CAE application software, life and material sciences and software development. HP is also expanding its HPC Partner Software Suite to include new offerings from DataSynapse, PathScale, Portland Group, Scali and TurboWorX that are available from HP and its authorized resellers. "HP Cluster Platforms provide a solid reference configuration for Accelrys' Materials Studio applications such as CASTEP and DMol3, while offering flexible solutions to meet a diverse set of customer needs," said Keith Glassford, director, Materials Science Marketing, Accelrys. "HP's industry-standard Cluster Platforms together with HP-MPI establishes a well-defined solution that we can confidently recommend to our users." Scalable visualization Later this year, HP plans to add a high-end scalable visualization solution that will complete the Unified Cluster Portfolio's integration of computation, data management and visualization in a single, integrated cluster environment. The solution will add high-performance HP Workstations into HP Cluster Platform configurations that combine with industry-standard visualization components and integrate with the HP XC System Software and HP SFS. The tight integration of scalable computation, data management and visualization will enable visualization of large data sets and dynamic data with high frame rates, as well as interaction with running simulations. More information on HP SFS 2.0 is available at www.hp.com/techservers/products/sfs.html; more information on the HP Unified Cluster Portfolio is available at www.hp.com/go/clusterportfolio2.