With the latest injection of funding from the National Science Foundation for a nationwide computing system, researchers like Rick Stevens and Ross Overbeek should have an easier time building up resources and users for the four-year-old TeraGrid project.
NSF recently awarded a total of $150 million over five years to operate and enhance TeraGrid, a compute system dedicated to open scientific research. Outside of the US Department of Energy’s classified weapons lab systems, TeraGrid is the world’s largest computing, storage, and networking system.
More formally known as the Extensible Terascale Facility, TeraGrid launched in 2001 with an initial NSF award of $53 million. The system was built up and went into full operation a year ago. The grid, which features more than 60 teraflops of computing power, a petabyte of storage, and an optical network speed of up to 40 gigabytes per second, has already been put to work on questions in genomics and disease analysis. NSF says it expects the current group of 1,000 users to grow to as many as 4,000 by 2009, the year funding runs out on the current award.
About a third of the new money — $48 million in total — will go to the University of Chicago, which coordinates TeraGrid through its Grid Infrastructure Group. Charlie Catlett, project director, says the money will go toward overall architecture, software integration, operations, and the coordination of user support. The money will fund “GIG activity with partnerships with other sites as well,” Catlett adds.
The remaining $100 million is allocated to resource provider sites that participated in TeraGrid’s construction and continue to support services and resources for it. Those sites include Indiana University, the National Center for Supercomputing Applications, Pittsburgh Supercomputing Center, Purdue University, San Diego Supercomputer Center, Texas Advanced Computing Center, and the University of Chicago/Argonne National Laboratory.
Among the new efforts for TeraGrid are 10 so-called Science Gateways aimed at getting new groups of researchers to use the compute resource. This community-tailored access will be provided in the form of Web portals, desktop applications, or via other grids.
Two of the gateways revolve around bioscience. In one, Daniel Reed at the University of North Carolina is spearheading a project to build biological communities by linking them in to TeraGrid. His proposal takes advantage of grid architecture to foster collaborations and allow groups to share diverse resources in unified environments. For the other initiative, Rick Stevens at the University of Chicago and Ross Overbeek of the Fellowship for Interpretation of Genomes have proposed a Web-based gateway to all data resources supporting research in microbial pathogens. Their National Microbial Pathogen Data Resource Center will be rooted in a well-annotated set of genome sequences, and also feature integration with related datasets on proteomes, transcriptomes, and metabolomes.
— Jennifer Crebs