[an error occurred while processing this directive]

NERSC 3 Greenbook

Next: The Nature of ER Up: DOE/OCTR/MICS White Paper Previous: DOE/OCTR/MICS White Paper

Conclusion and Strategy from the discussion below:

The target heterogeneous distributed computing system for Energy Research would satisfy the capability and the capacity needs for Energy Research investigators. However, the target computing system may well not be attainable in the present outlook on the DOE budgets for the remainder of the century. A plan with less risk is to make the access center of highest priority because of the flexibility of the high performance computing and storage system and the strongest ability to accomplish the major computational tasks of Energy Research. Provided there is sufficient data communications technology, the centralized facility can tradeoff capability computing for capacity computing while the distributed system of workstations and symmetric multiprocessing units without the access center cannot provide capability computing. Moving away from the access center in the upcoming tight budget times will also place an additional load on the ER science and technology offices to provide local computing services when they can least afford them. This approach also depends less on technologies to be developed and progresses naturally toward the perceived nature of computing in the 21st century - heterogeneous distributed computing. Funds would be provided later for powerful workstations as their capability improves and their price comes down.

NERSC 3 Greenbook

Next: The Nature of ER Up: DOE/OCTR/MICS White Paper Previous: DOE/OCTR/MICS White Paper