CMS Data Grid System Overview and Requirements

This document gives a comprehensive overview of the data grid system that CMS intends to operate around December 2003. This CMS data grid system will support both CMS production and analysis and will seamlessly tie together resources at CERN and at international CMS regional centers. This document focuses on the relation between the CMS software components and the grid software components operating inside the 2003 CMS data grid system. In addition, the document includes overview and reference material to introduce the members of the grid projects (GriPhyN, PPDG, and the EU DataGrid) to the CMS data handling environment. This document contains a snapshot, taken in 2001, of the vision that CMS has of the intended software capabilities of its production data grid system in 2003, and the expected scaling towards 2007. To capture the expected level of complexity, the vision is sometime s worked out to considerable detail, even though some of these details are likely to be adjusted in future. Though the vision captured in this document will likely evolve, this document does yield the current requirements for the grid projects that CMS is involved in as a `customer', the requirements for the grid components which are to be created from now until the end of 2003. The major CMS software milestones affecting the grid projects are the `delivery of baseline core software' milestone for December 2002, this includes the choice and integration of grid components into the baseline software, and the `20% data challenge' milestone, for which the work starts in January 2004 with milestone completion in December 2004. The 20% data challenge includes the full range of distributed operations required for the analysis of CMS data under realistic conditions as occurring during LHC ope ration from 2006 onwards. The primary audience for this document are the participants in the grid projects that CMS is involved in.