The program, called Scientific Cluster Support, is run by the UNIX Systems Group in the Information Technologies and Services Division. The program helps selected research groups with choosing the right computer configuration, setting up and configuring the system, helping researchers get their applications onto the cluster, then providing ongoing support in a room equipped with the necessary networking, cooling and electrical systems, as well as appropriate cybersecurity measures.

Computer clusters consist of a number of commodity processors connected with a specialized switch and software to perform as a single system, providing researchers with cost-effective high-performance computing. Although often viewed as “plug and play” systems, clusters are complex and usually require specialized expertise to configure and operate at peak performance.

The Laboratory-funded, four-year Scientific Cluster Support program plans to provide the necessary support and service to 10 clusters being purchased by groups in seven scientific divisions. The clusters will be phased into the program over the next two years.

“In addition to promoting the use of Linux-based clusters for scientific computing, we are providing the necessary support to ensure that these systems can be utilized most effectively,” said Gary Jung, project manager. “Ultimately, our goal is to allow scientists to best spend their time on research and let us handle the aspects of supporting their cluster.”

The first cluster, an 84-processor system purchased by a group led by Arup Chakraborty of Physical Biosciences and Materials Sciences, has been installed in the Bldg. 50B computer facility and is now in production. Chakraborty's group said they will use the cluster to advance two major projects in fundamental biology and bio-sensor technology. The two projects are: (1) how receptor-ligand binding, cell membrane properties, and intracellular signaling conspire together to effect T cell recognition of antigen in the adaptive immune system, and 2.) how DNA hybridization and protein-protein binding can be translated into a mechanical signal for pathogen detection.

Other clusters scheduled to be added to the support program this year include a cluster arriving this week to be used by Ashok Gadgil and Shaheen Tonse of the Environmental Energy Technologies Division, as well as systems for Priscilla Cooper in Life Sciences, Michael Eisen in Life Sciences, Martin Head-Gordon in Chemical Sciences, and William Miller in Chemical Sciences. In the second year, clusters already being used will be brought into the support program. These are systems being used by Sung-Ho Kim, Steven Brenner, Paul Adams and Stephen Holbrook in Physical Biosciences, Bill Lester in Chemical Sciences, Martin White in Physics, and Mike Hoversten and Ernest Majer in Earth Sciences.

The Scientific Cluster Support program grew out of a Lab-wide assessment of the best approach to increase mid-range computing in Lab research projects. The assessment was conducted by the Midrange Computing Working Group of the Computing and Communications Services Advisory Committee, or CSAC. At a Midrange Computing workshop held a year ago, Lab scientists identified a comprehensive cluster support program as the best path forward to increase mid-range computing at the Lab.

The SCS project team consists of Tammy Welcome, Gary Jung, Greg Kurtzer, James Buszard-Welcher, Jackie Scoggins and Susan James.