Long-Term Massive Production Runs on Alliance Resources: Experience

V. Litvin, H. Newman, S. Schevchenko

California Institute of Technology

During 2003 Caltech team is producing and analysing 10M of events for further study Higgs->gamma gamma channel with full background simulation based on cmsim125 and ORCA6 software. All backgrounds have been improved ten-fold compare to the previous study. Physics results has been presented. Alliance resources have been heavily used for that task. TeraGrid has been used as well. cmsim has been ported on Itanium II IA64 architecture for that purposes. The main problem for the long term (over month) runs is the robustness of clusters and Mass Storage Systems as well as some restrictions coming from recent computing policies on NCSA clusters. Some statistics of cluster's usage has been presented.