SCIENCE

King of Prussia, PA, and Milpitas, CA – InfiniCon Systems, the premier provider of shared I/O and switching solutions for next-generation server networks, and Appro, a leading developer of high-density, high-performance servers, today announced they are demonstrating at the ClusterWorld Conference and Expo (San Jose, June 24-26) the first InfiniBand-based High Performance Computing (HPC) server cluster to incorporate both Dual AMD Opteron and Dual Intel Xeon processor technology. The HPC solution (Appro Booth # 713) consists of Appro’s HyperBlade Server Cluster, which supports up to 80 compute blades/160 CPUs in one cabinet; InfiniCon’s InfinIO 7000 Shared I/O and Clustering System, which provides servers with a standards-based, 10 Gbps InfiniBand interconnect; and InfiniCon’s InfiniServ 7000 Host Channel Adapter (HCA), which enables server-to-server communications via the Message Passing Interface (MPI) protocol. The cluster will utilize the Linux Red Hat 7.3 operating system and run a graphics-rendering application. Affordable, Scalable Computing Power from Industry Standard Technology Facing flat IT budgets and driven by the directive to do more with existing resources and staff, enterprise data center managers have been drawn to the superior bandwidth and low-latency advantages that the InfiniBand Architecture provides for building low-cost HPC clusters from commodity hardware and standards-based networking technology. HPC cluster configurations based on InfiniCon and Appro technology make it possible to add server nodes incrementally to an ultra-scalable InfiniBand fabric with plug-and-play ease, enabling users to provision compute power as needed to support data-intensive HPC applications. The InfinIO 7000 also provides shared I/O gateways that Appro/InfiniCon HPC server clusters can use for seamless integration into existing Fibre Channel and Ethernet networks. "HPC environments tend to require large volumes of servers, which creates on-going management and budget challenges," noted Vernon Turner, Group Vice President, IDC Enterprise Server Solutions. "The integrated offerings from InfiniCon and Appro are addressing increasingly tactical deployment issues in the HPC market today like server density, heterogeneous platform support, and the capability to provide clusters with common access to storage and network resources. This combination merits strong consideration from data center managers striving to drive down the cost of computing while increasing the scale of its capabilities." Contrasted to older, proprietary HPC architectures which run at only 1 to 3 Gbps, the InfiniBand-based cluster demonstrated by InfiniCon and Appro at ClusterWorld affords full 10 Gbps links for system-to-system communications, with latency levels measured at less than 7 microseconds – a key metric for assessing how efficiently application tasks are executed across a fabric. "Our strategic relationship with InfiniCon enables Appro to offer a totally managed HPC solution that can incorporate any of our full range of high-performance server platforms and technologies, including the new 64-bit processing platforms," noted Daniel Kim, president and CEO of Appro. "Customers can build very cost-effective, high-bandwidth clusters today on an infrastructure that is flexible and easy to scale."