NSF CISE Grand Challenges in e-Science Workshop
Participants: Jason Leigh, Maxine Brown, Thomas A. DeFanti, Joe Mambretti
Institutions: National Science Foundation
This Workshop will answer the question: Do the Grand Challenge Applications in e-Science need an experimental network? If yes, why? And, what should it be in the near and longer term?
In 1992, NSF 92-7, a solicitation for Grand Challenge Application Groups, was issued. It specifically provided funding for multi-disciplinary groups of scientists, engineers and mathematicians to apply emerging high-performance computing and communications systems to advance the solution of diverse science and engineering problems. Activities supported under this announcement were “expected to achieve significant progress on Grand Challenge Applications fundamental problems in science and engineering, with broad economic and scientific impact, whose solution could be advanced by applying high performance computing techniques and resources.” At that moment in time, the NSF supercomputer centers were in full swing, but the Internet was all of 45Mb and just barely being used.
Almost ten years later, it is time to revisit the Grand Challenges approach, but this time focusing on high-performance networking and cluster computing. Networking capacities on single channels now are 10x faster than the computers attached to them, arguably leading to a paradigm shift. The question this Workshop needs to answer is: Do e-Science applications need a many-10Gb network of their own?
If the answer is yes, and justified by example, then the implementation questions need to be answered. Can NSF afford to build a National network once more? Should it? Or, rather, can the research be done on much cheaper multi-10Gb metropolitan and regional networks for the next few years? Is there a need to support international connectivity at 10Gb or higher? Should NSF heavily fund the applications and let the PIs procure the bandwidth, middleware, and toolkit support necessary, or should multi-science testbeds featuring multi-10Gb bandwidth be funded with applications teams supported to use them?
This Workshop will result in the publication of a report advising NSF on a funding program for Grand Challenges in e-Science. The report will contain a list of Recommendations with justification. All participants? names will appear as contributors to the resultant document. DeFanti and Brown will be the editors of the report.
This Workshop has been organized at the request of the NSF CISE Advanced Networking Infrastructure & Research (ANIR) Division.
Date: December 5, 2001 - December 6, 2001