Prof. Dr. Kathy Yelick, Director of NERSC, Associate Berkeley Lab Director for Computing Sciences, & Professor of Computer Science, Lawrence Berkeley National Laboratory (LBNL) & University of California at Berkeley, USA

Cloud computing has proven to be an attractive modeling for business computing, and several scientific teams are also experimenting with clouds for their scientific computations. The NERSC Center at Lawrence Berkeley National Laboratory has constructed a testbed to explore scientific cloud computing as part of a project called Magellan, a joint effort with Argonne National Laboratory. Building on the NERSC experience serving a broad scientific workload with over 3000 users, this talk will examine some of the characteristics of scientific applications, including both data and compute-intensive problems, and the suitability of the cloud computing model. Magellan is specifically targeting small- to medium-scale serial and parallel applications as a possible complement to petascale systems that are used for the highest end computing. The talk will examine questions of how scientific clouds should be configured from a hardware and software standpoint, consider the value of some of the cloud programming models such as Map-Reduce, look at how the cloud business model translates to a scientific environment, and report on some of the early successes with Magellan for bioinformatics applications.