Week 1 Summary From David Stickland
On Monday March 1 the CMS Data Challenge DC04 got formally underway; it will run for two months.
The main goal is to run for sustained period at 25Hz input rate; roughly 1/4 of the expected LHC startup conditions. The data are reconstructed at CERN, distributed to the T1 centers for further processing and eventually on to T2 centers for some specific analysis activities.
The initial phases of the challenge involve CERN as the Tier-0 Computing center, and Tier-1 centers at: RAL-UK, CNAF-Italy, PIC-Spain, IN2P3-France, GridKA-Germany and FNAL-USA. Tier-2 centers are gearing up to participate later.
The first five days have seen remarkable success and progress. Reconstruction runs at the CERN T0 have run 600,000 events without any failures. The jobs are running in reasonable time (20-60s/ev) and memory footprint. The reco jobs produce our first DST events containing for example fully reconstructed tracker, muons trigger results, calorimetric clusters, jets etc. This data is passed to distribution buffers at CERN where we are now exercising the automatic catalog construction and safe transfer of the DST out to all the participating T1 centers and their tape vaults. A very large number of software agents and databases are being debugged and brought into operation
We have the goal of finalizing the ORCA-reconstruction code next week and ramping up full production about next weekend. PRS groups and T1 centers are preparing pseudo-analysis tasks to demonstrate the ability to further process the DST as it arrives in the T1 centers. We hope to eventually demonstrate the ability to run analysis tasks on the DST using all the LCG computing resources that can have access to the data copies.
Thanks to everyone for your sustained hard work!