---

Computing Technical Design Report

5.2 Relations with the LCG Project and with Grid Middleware Providers

Computing resources can be accessed by ATLAS physicists through Grid middleware components and services. The principles and objectives of computational data Grid technology aim to provide capabilities that are well suited to the distributed operation foreseen by the ATLAS Computing Model. These include services for software installation and publication, production operations and data access for analysis through a uniform security and authorization infrastructure, interfaces for remote job submission and data retrieval, and job scheduling tools designed to optimize utilization of computing resources.

The LHC Computing Grid (LCG) Project [5-1] was set up at the beginning of 2002 to deploy and operate Grid middleware for the LHC experiments, and to help them to integrate Grid tools with their software base. Currently the LCG project has a dual role:

1. it acts as an umbrella organization for all Grid middleware deployments that are relevant for LHC experiments;
2. it deploys and operates one particular middleware suite (LCG-2 and later updates) in cooperation with the European Union funded EGEE middleware project.

Three "flavours" of Grid middleware are deployed on computing resources that are available to ATLAS: Grid3/OSG in the USA, NorduGrid/ARC in Scandinavia and a few other countries, and LCG-2/EGEE in most of Europe, in Canada and in the Far East. In order to make optimal use of all resources, the ATLAS production and analysis systems are designed to be independent of any particular Grid flavour; only the few software components that access directly specific Grid tools are Grid-flavour dependent.

However, as ATLAS computing must work across the Grid deployments in a coherent and sustainable way, it is vital that:

• the Grid deployments used by HEP ensure the highest possible degree of interoperability at the service and API level;
• that changes in those services and APIs be advertized well in advance, and technical support be made available to experiment developers where applicable.

It is expected that a major focus of the LHC Computing Grid project will be in promoting the interoperability between the various Grids and encouraging convergence wherever possible. It will also have a role in developing a common layer above the various Grid deployments if that is needed. At present, such layered tools are being developed within the experiments, but common tools developed in a timely fashion would enhance sustainability.

ATLAS has strong representation in the LCG organization, and participates actively in the monthly meetings of the LCG Grid Deployment Board and the weekly operations meetings organized by the Grid Deployment Area. For example, the ATLAS Computing Coordinator sits on the LCG Project Execution Board. ATLAS collaboration members sit also on the management boards of the three Grid middleware development projects, and guarantee a constant flow of information in both directions.

Copyright © CERN 2005