Transition Costs Associated with the NCAR-Wyoming Supercomputer Center (NWSC) Facility

University Corporation For Atmospheric Res, Boulder CO

Investigators

Abstract

The Computational and Information Systems Laboratory (CISL) at NCAR will transfer its principal supercomputing operations from the NCAR Mesa Laboratory (ML) in Boulder, Colorado to the newly-constructed NCAR-Wyoming Supercomputer Center (NWSC), located near Cheyenne, Wyoming. The goals of this project are to maximize scientific productivity and efficiency during the transition, seamlessly integrate NWSC into NCAR's existing cyberinfrastructure (CI) and the broader NSF CI, and significantly increase the service and security of all supercomputing operations that NCAR provides for the research community. Meeting these goals requires careful and detailed planning by NCAR and CISL staff. To avoid disruptions, certain services will have to be duplicated at the NWSC and ML facilities during the transition. To hold down costs, NCAR plans to relocate some ML equipment to NWSC, while other equipment will continue in service at ML or be decommissioned. Activities funded under this award include staffing the new center and relocating equipment to Cheyenne. In addition, facility monitoring software will be purchased that would enable both the Boulder and Cheyenne facilities to be operated by a single team in Wyoming. The remainder of the award would fund the additional utility costs of running both systems concurrently during the transition and the purchase of servers and related equipment needed for the transfer of NCAR's considerable data holdings (~10s of petabytes) to the NWSC. A smooth, well-executed transition to the NWSC will help to ensure uninterrupted support for the scientific community that depends upon this crucial computing facility. The demand for NCAR's current computing facility demonstrates the community's urgent need for petascale computers and associated cyberinfrastructure. A seamless operational transition will support this scientific momentum by enabling investigators to achieve dramatic improvements in model resolution, better physical process representations, longer simulation lengths, and better statistics for a wide range of important Earth System science applications.