Lawrence Berkeley National Laboratory is one of the nation’s primary centers for cloud computing adoption and research and is the site of a number of experimental forays into the world of virtualized HPC applications . The lab is in the process of implementing Google Apps and has been the site of extensive testing of Amazon’s EC2 and more recently, the HPC-oriented Cluster Compute Instance type (CCI) as well as the site of the large Magellan private cloud project.
Berkeley Lab, which operates under the authority of the Department of Energy (DOE), has Alvarez “on loan” from the agency, The CIO spends a small part of the month on site at the research center before heading back to her home base in D.C. where she leads several strategic initiatives within the DOE. Part of what Alvarez has been tasked with is looking at Berkeley Labs’ current HPC resources, which include a number of supercomputers, smaller clusters and its Magellan cloud, to determine how such resources might be optimized and become more available to researchers.
This week John Foley at Information Week talked with Dr. Alvarez about the range of HPC and cloud projects underway in Berkeley, including their use of Amazon’s Cluster Compute Instances. According to Alvarez, the center has “been testing that in terms of performance and scalability to see if we can offload some of our demand during peak workloads” and that their “initial results show that performance and scalability are comparable to what we can provide in our data center on our hardware.” She notes that while this has played out successfully, it now becomes a question of cost analysis.
Dr. Alvarez told Foley that when she migrates back to Washington, part of her joint duties involve working with the White House and Office of Management and Budget to develop “a policy for handling federal requirements and guidelines that com to the department, assessing their scope and availability.”