Computing at the Large Hadron Collider: Extracting Physics from 40 Million Collisions a Second

The Large Hadron Collider (LHC), in Geneva, Switzerland, has recently begun operation at the highest energy ever achieved by a particle accelerator. While the tens of petabytes of data which will be produced within the first year of running will provide a fertile ground for testing many physical theories, recording, distributing, and analyzing this data will prove to be a difficult computational science problem. This talk, designed for non-particle physicists, will provide an overview of the computational challenges faced by experiments at the LHC. In particular, I will focus on the Trigger - a hybrid hardware/software system designed to filter data in real time from a collision rate of 40 MHz to a write to tape rate of 200 Hz - and the Grid - a software system for distributing data and jobs to the roughly 20k compute nodes used for analysis of LHC data at institutions and universities around the world.