The ATLAS Computing Model

The ATLAS computing model was constructed to exploit the opportunities of the Grid in handling the large volumes of data from the Large Hadron Collider and to allow easy and relatively local access to data for all of its collaborators worldwide. Despite delays with collision data, the model has now been tested with beam-related and cosmic ray data, and much has been learned. The model has retained its overall design, but with adjustments for the actual functionality of the delivered Grid middleware and services, and for the realities of data access. While much has still to be learned, the model has worked effectively. This paper covers the roles of the various Tiers, the resources required and the expected workflows.