R. J. Aiken, "The Grand Challenge of Managing the Petascale Facility," Argonne Report ANL/MCS-07/5, December 2006. [pdf]
This report is the result of a study of networks and how they may need to evolve to support petascale leadership computing and science. As Dr. Ray Orbach, director of the Department of Energy’s Office of Science, says in the spring 2006 issue of SciDAC Review, “One remarkable example of growth in unexpected directions has been in high-end computation.” In the same article Dr. Michael Strayer states, “Moore’s law suggests that before the end of the next cycle of SciDAC, we shall see petaflop computers.” Given the Office of Science’s strong leadership and support for petascale computing and facilities, we should expect to see petaflop computers in operation in support of science before the end of the decade, and DOE/SC Advanced Scientific Computing Research programs are focused on making this a reality.
This study took its lead from this strong focus on petascale computing and the networks required to support such facilities, but it grew to include almost all aspects of the DOE/SC petascale computational and experimental science facilities, all of which will face daunting challenges in managing and analyzing the voluminous amounts of data expected. In addition, trends indicate the increased coupling of unique experimental facilities with computational facilities, along with the integration of multidisciplinary datasets and high-end computing with data-intensive computing; and we can expect these trends to continue at the petascale level and beyond. Coupled with recent technology trends, they clearly indicate the need for including capability petascale storage, networks, and experiments, as well as collaboration tools and programming environments, as integral components of the Office of Science’s petascale capability metafacility.
The objective of this report is to recommend a new cross-cutting program to support the management of petascale science and infrastructure. The appendices of the report document current and projected DOE computation facilities, science trends, and technology trends, whose combined impact can affect the manageability and stewardship of DOE’s petascale facilities.
This report is not meant to be all-inclusive. Rather, the facilities, science projects, and research topics presented are to be considered examples to clarify a point.