Expand the Capability and Capacity of NCAR Supercomputing Facilities

This diagram shows the rate at which heat generated by each type of equipment in a data center has increased over the last 14 years, and it projects how heat loads will rise in the near future. When these heat loads are added together, it is clear that a data center's cooling capacity is critical to its ability to grow. Existing data centers will be unable to cope with the heat loads of next-generation computing equipment, and new mechanical and electrical infrastructure will need to be built.

NCAR is at both a challenging and exciting time with respect to providing facilities to operate computing and related cyberinfrastructure. The continued trend of higher power density coupled with the ability to purchase more computer equipment for nearly the same costs year to year, strain all existing facilities. This trend appears to continue with projections into the 2014 time frame, as shown in the diagram. The diagram is from Datacom Equipment Power Trends and Cooling Applications, 2005, ASHRAE Special Publication, 112 pages.

Meeting the challenge of providing facilities sufficient for NCAR is fundamental to NCAR's strategic goal to "Provide robust, accessible, and innovative information services and tools," and to the related strategic priority of "Enhancing capability and capacity of NCAR supercomputing." We have begun a two-pronged approach to meeting this challenge.

First, we have solidified and prepared the NCAR Mesa Lab computing facility infrastructure to operate at or near capacity for the next three years. This included completing a risk assessment and implementing the recommendations over the last two years. When the ICESS equipment is installed in early FY 2007, the Mesa Lab will be operating at its maximum capacity.

Second, we are continuing to develop plans to construct a new facility. These plans evolved rapidly over the past year. Partnership discussions were held with a number of institutions, and potential sites were evaluated for their technical feasibility. In part because of these discussions, excitement for a potential large geoscience center arose. As a result, a workshop on High Performance Computing for the Geosciences was planned and completed at the end of FY 2006. In parallel, work continues to move from NCAR's existing conceptual design toward a construction design. Both efforts will continue into FY 2007.

This work is supported by NSF Core funding.