A second large responsibility is assisting Run II and other experiments. Aside from ongoing consulting and design/code review activities, one advance is the introduction of a significant set of performance study tools with which we will be helping experiments to optimize there processing usage. This has already started paying dividends; we have been able to find some enddemic costly techniques in D0 code, and to help a CDF users find and eliminate causes for a factor of five decrease in performance. The tools stemmed from brainstorming for a "professional development" session on performance tools.
One member of this group has more direct responsibility for CDF software; that experiment reccognizes her value and has in fact made her co-head of CDF offline software. Another member has responsibilities for leading the RTES multi-institutional developments; he ahs begun indroducing a "use-cases" methodology to improve the prospects of diverse academic groups producing a coherent and practical overall product. Several members are involved in projects to assist the Beams Division in orbit-smoothing, shot data analysis, and related matters. Some contributions to the SDA work are in place and useful; the orbit-smoothing work has gone slowly, due partly to conflicting time requirements related to the new goals assessments, and partly to the need to acquire a deeper understanding of the beams physics involved. Finally, beginning early in FY 04, there is likely to be an assignment of the equivalent of 1 FTE to CMS, to do software development under their direction.
Two members of the SLD group are on assignment directed by US CMS on a half-time basis. They are participating in efficiency studies and in evaluation of designs for large library code collections.
In addition the group works closely with the experiments to provide documentation and code for client applications that use the database information. Work has largely centered on Oracle applications, but the goal is to provide general interfaces that allow for multiple commercial and freeware database solutions as well. The group also provides a monitoring framework and tools that allow the database delivery systems to be alarmed, monitored for performance, and diagnosed for problems. This set of tools came about because of realization by several people across the whole section that multiple efforts to improve monitoring could benefit from a reasoned, coherent shared effort. The monitoring tools came to fruition in less than a year, and are paying dividends. For example, there is now daily identification of each experiment's most heavy users; by working with them to reduce inefficiencies, we have brought database crises under control.
Recently, the group has been helping to review parameter databases being used in the beams division to track magnet and other information for Tevatron operation. This appears to be a project that will continue for the next several weeks and we hope will play an important role in improving the operation of the accelerator. We are currently making recommendations concerning moving some of their existing data to a new, more maintainable database, unifying into one modern database product such as SYbase, PostgreSQL, or Oracle. This database would be prepared for the influx of new survey data starting in mid-fall.
September 14, 2003