Using CAS to Manage Role-Based VO Sub-Groups

LHC-era HENP experiments will generate unprecidented volumes of data and require commensurately large compute resources. These resources are larger than can be marshalled at any one site within the community. Production reconstruction, analysis, and simulation will need to take maximum advantage of these distributed computing and storage resources using the new capabilities offered by the Grid computing paradigm. Since large-scale, coordinated Grid computing involves user access across many Regional Centers and national and funding boundaries, one of the most crucial aspects of Grid computing is that of user authentication and authorization. While projects such as the DOE Grids CA have gone a long way to solving the problem of distributed authentication, the authorization problem is still largely open. We have developed and tested a prototype VO-Role management system using the Community Authorization Service (CAS) from the Globus project. CAS allows for a flexible definition of resources. In this protoype we define a role as a resource within the CAS database and assign individuals in the VO access to that resource to indicate their ability to assert the role. The access of an individual to this VO-Role resource is then an annotation of the user's CAS proxy certificate. This annotation is then used by the local resource managers to authorize access to local compute and storage resources at a granularity which is base on neither VOs nor individuals. We report here on the configuration details for the CAS database and the Globus Gatekeeper and on how this general approch could be formalized and extended to meet the clear needs of LHC experiments using the Grid.