NPACI ROCKS Simplifies Deployment of INTEL Itanium-Based Linux Clusters

Published 06/11/2002

SAN DIEGO, June 11, 2001 - Developers from the San Diego Supercomputer Center (SDSC) at the University of California, San Diego (UCSD), have released NPACI Rocks 2.2.1, which extends the easy-to-use open-source cluster toolkit to Linux clusters powered by Intel Itanium processors. Clusters provide a stable, standardized platform for academic research and industrial uses such as financial and multimedia services. Running NPACI Rocks on clusters with the new 64-bit Intel Itanium processors gives users access to features capable of processing huge amounts of data, managing peak workloads, and transactions of various sizes, while maintaining security, responsiveness, availability, and reliability. The Itanium processor architecture provides an unprecedented capability to store, deliver, and mine data.

"We've been driven by one goal: Make clusters easy‹easy to deploy, manage, upgrade, and scale. Rocks is staying ahead of the technology curve by supporting Itanium-based clusters, one of the highest-performance computing environments available to applications scientists. Making clusters easy also means that we must provide a rational path through these types of technology transitions," said SDSC's Philip Papadopoulos, leader of the NPACI Rocks development effort.

Researchers using clusters running NPACI Rocks have recently achieved a number of notable scientific successes, including landmark resolution imaging of biomolecules, such as microtubules, simulations of the junctions between nerve cells, and novel techniques for matching potential drugs with molecular targets.

Robert Konecny, Director of the Keck II Computational Center at UCSD, which has over 100 users investigating projects in biology, chemistry and physics on its 76-processor cluster said, "Rocks is an excellent clustering solution which makes the building of powerful computational clusters extremely easy. The thoughtful Rocks software design hides the complexity of cluster building and management for casual users, while providing sophisticated and very powerful tools for more advanced deployment."

Detlef Stammer, an associate professor of physical oceanography at the Scripps Institution of Oceanography at UCSD, uses two 256-processor NPACI Rocks-managed clusters for ocean modeling and ocean state estimation. "Rocks has several advantages," Stammer said. "The first one is that we can work very closely with the Rocks development group to tailor it to our needs. The other is that the software manages the entire cluster in an easy, user-friendly way."

Ian Fisk, a UCSD physics researcher involved with the international Compact Muon Solenoid experiment, supports clusters with NPACI Rocks. "Rocks significantly reduces the human effort required to manage our cluster," Fisk said. "The ability to easily change the operating system version and configuration on our computational nodes allows us to efficiently serve as a computing center for several experiments."

TECHNICAL DETAILS

NPACI Rocks adds functionality to the base Red Hat 7.2 Linux distribution without specific kernel hooks. This approach allows the software to handle the natural evolution of Linux updates. It enhances the Linux cluster environment with features that allow users to start, observe, and control processes on cluster nodes from the cluster's front-end computer while supporting standard Linux interfaces and tools. The result is a stable and extensible environment that appeals to both end users and software developers, and provides a supported platform for the deployment of advanced clustering applications.

NPACI Rocks provides turnkey software installation and update for Linux clusters, as well as cluster tools such as the Portable Batch System (PBS), Maui Scheduler, and MPICH for Ethernet and Myrinet. Features new to NPACI Rocks 2.2.1 include:

• Upgraded Ganglia server and client, used for collecting and visualizing cluster-wide monitoring metrics
• Upgraded MPICH-GM implementation for Myrinet interconnects
• The "stream" memory bandwidth benchmark
• Pre-packaged ATLAS BLAS and HPLinpack (from the University of Tennessee)
• Parallel Virtual File System (PVFS) support

NPACI Rocks was originally developed at SDSC and partners at the University of California, Berkeley. An additional site of the NPACI Rocks development team in Singapore provides support for PVFS. More about NPACI Rocks, including downloads, can be found at http://rocks.npaci.edu/.

The San Diego Supercomputer Center (SDSC) is an organized research unit of UCSD and the leading-edge site of the National Partnership for Advanced Computational Infrastructure (NPACI). SDSC's mission is to develop and use technology to advance science, and SDSC provides leadership both nationally and internationally in computing, data management, biosciences, and other areas. As a national laboratory for computational science and engineering, SDSC is funded by the National Science Foundation through NPACI and other federal agencies, the State and University of California, and private organizations. For more information, see www.sdsc.edu or contact David L. Hart, SDSC Communications, 858-534-8314, dhart@sdsc.edu.