[ Up ] [ Current Status ] [ Previous Status ]

Computational Resource Status - Q3 1999

On August 3, 1999 Cornell University announced installation of a 256-processor cluster at the Cornell Theory Center (CTC). The system consists of 64 Dell PowerEdge servers, each with four Intel Pentium® III Xeon 500 Mhz processors and running the Microsoft Windows® NT operating system. Each system has 2 MB of Level 2 cache per processor, 4 GB RAM and 54 GB disc. The primary cluster interconnect is provided by Giganet, Inc. Dubbed AC³ Velocity, the cluster will serve as a production high-performance computing resource for CTC's research community and the Advanced Cluster Computing Consortium (AC³).

AC³ is a Cornell research and IT service consortium established for corporate, higher-education, and government agencies interested in the effective planning, implementation, and performance of commodity-based software, systems, and tools.

Dell, Intel, Microsoft Corp., and Giganet, who are infrastructure members of AC³, view the consortium as an important step in accelerating the standardization of cluster computing.

"Dell is committed to building standards-based solutions that scale throughout the enterprise," said Michael Lambert, senior vice president of Dell's enterprise systems group. "We applaud the teamwork that led to creation of AC³ Velocity and look forward to working with AC³ members to further validate and demonstrate the computing power, availability, and scalability of Dell PowerEdge servers and PowerVault storage solutions in enterprise environments."

"The emergence of the Internet is changing the face of contemporary business and scientific computing, increasing the demand for server scalability and availability, while keeping costs under control," said Justin R. Rattner, Intel Fellow and director of Intel's Server Architecture Lab. "Clusters of servers, based on Intel Pentium® III Xeon processors and multiple industry-standards, like Virtual Interface Architecture, have emerged as one of the premier strategies to provide both high availability and scalability in the computationally-intensive Internet environment."

Todd Needham, manager of research programs at Microsoft, added, "We're focused on working with the AC³ to build and validate a model for supercomputing constructed from industry standard servers, high-performance networking, and Windows NT that industry and large enterprises can apply to a wide range of problems." CTC will rapidly be moving toward Windows 2000 cluster services on AC³ Velocity this year. Needham commented, "At 256 processors, Velocity will clearly be a testament to the scalability of Windows 2000."

"Today's environments require enterprises to be able to easily scale computing needs," said Gareth Taube, Giganet's vice president, marketing. "With the combination of our cLAN products, powered by Dell's PowerEdge servers, we can provide a flexible, cost-effective way to support mission-critical scientific, engineering, and Internet applications." Taube added that AC³ Velocity will be the largest Giganet production configuration in the world.

"We are confident that AC³ Velocity will deliver scalable, high-performance computing with enterprise-wide applicability," said David Lifka, systems director at CTC. Thomas F. Coleman, CTC director, noted that researchers at Cornell, one of the leading institutions in computational science and engineering in the world, agree. "We're finding that Windows NT-based cluster computing is an attractive environment for computer scientists and computational scientists, and is bringing new researchers from business and the social sciences into the CTC community," he said.

The AC³ provides access to both pre-release hardware and software, as well high-performance production machines. AC³ Velocity will be used for technical applications such as biomedical and genomics research, seismic processing, materials modeling, large-scale database and datawarehousing applications, and computer science research in areas such as parallel I/O and systems reliability.

Researchers at Cornell's Johnson Graduate School of Management applauded CTC's move to Microsoft Windows NT and Microsoft Windows 2000-based cluster computing. Andrew Ainslie, for example, conducts marketing research using very large datasets and computationally-intensive techniques. "The Windows NT-based supercomputing environment presents researchers with a tremendous opportunity," he said. "We are able to develop code on inexpensive machines in our offices, and then scale the application massively when it is ready to run." According to Ainslie, this provides an excellent, highly flexible framework for his research, enabling him to tackle projects "at least one order of magnitude larger than I would have been able to otherwise."

Ken Birman, Cornell computer scientist, is using the cluster to develop a new generation of software for fault-tolerant, load-balanced "data center" applications, such as servers managing massive databases. He commented that "Experiments using the new cluster will take our research from a conceptual stage to practical solutions that can be used in realistic settings. Cluster computing is poised to revolutionize many kinds of business applications, and with the AC³ Velocity, Cornell's contributions can have an immediate impact on potential users." Birman noted that a recent Presidential blue-ribbon committee, PITAC, cited scalable, fault-tolerant cluster computing as a critical requirement for this country over the next decade.

Cornell physicist James Sethna works on a challenging project focused on developing modeling and simulation capabilities to represent defect structure and evolution in solids over many length and time scales. According to Sethna, "The Theory Center is developing what may well become the working environment for affordable supercomputing of the future, making it easy to launch and control large-scale projects." Sethna noted that CTC's technical move to large-scale cluster computing meshes well with his group's Digital Material working environment, which will give researchers and engineers easy and flexible access to state-of-the-art mathematical and physical simulations and models."

Robert Constable, Cornell's new dean for computing and information sciences (CIS) believes that AC³ and AC³ Velocity will be valuable assets in Cornell's strategic vision for CIS in research and education. "Windows 2000 and Intel architecture-based cluster computing are the direction of the future," he said. "The Theory Center is an integral part of CIS and we're delighted that they are taking a leadership position in this arena."

In preparation for the arrival of the 256-processor cluster, CTC's Cluster Computing Solutions Group has been porting applications to a smaller cluster that was received as part of a Technology for Education 2000 grant that Intel made to Cornell in 1997. The grant enabled Cornell to develop expertise in moving applications from UNIX systems to Microsoft Windows NT, based on the Intel Xeon processor architecture, and provided excellent performance and scaling data. The group will provide suggestions for enhancing clustering technologies to AC³ Infrastructure Members based on the experiences of users with large-scale, computationally challenging problems.

AC³ Velocity will run MPI/Pro from MPI Software Technology, Inc. and implementations of OpenMP from Kuck and Associates and the Portand Group, with compilers for Fortran90, C and C++. MPI/Pro includes ClusterCoNTroller, a resource management and scheduling tool developed by Lifka at CTC and commercialized by MPI Software Technology. Other AC³ infrastructure members include Etnus, Inc., Fluent, Inc., ILOG, Inc., MPI Software Technology, Inc., the Numerical Algorithms Group, The Portland Group, Inc., Reliable Network Solutions Inc., SAS Institute, Inc., and Visual Numerics, Inc.

Last modified on: 10/07/99