R. Cavanaugh (UNIVERSITY OF FLORIDA)
A grid consists of high-end computational, storage, and network resources that, while known a priori, are dynamic with respect to activity and availability. Efficient co-scheduling of requests to use grid resources must adapt to this dynamic environment while meeting administrative policies. We discusses the necessary requirements of such a scheduler and introduce a distributed framework called SPHINX that schedules complex, data intensive High Energy Physics and Data Mining applications in a grid environment, respecting local and global policies along with a specified level of quality of service. The SPHINX design allows for a number of functional modules and/or distributed services to flexibly schedule workflows representing multiple applications on grids. We present experimental results for SPHINX that effectively utilize existing grid middleware such as monitoring and workflow management/execution systems. These results demonstrate that SPHINX can successfully schedule work across a large number of grid sites that are owned by multiple units in a virtual organization.