Status and Evolution of ATLAS Workload Management System PanDA

The ATLAS experiment at the LHC uses a sophisticated workload management system, PanDA, to provide access for thousands of physicists to distributed computing resources of unprecedented scale. This system has proved to be robust and scalable during three years of LHC operations. We describe the design and performance of PanDA in ATLAS. The features which make PanDA successful in ATLAS could be applicable to other exabyte scale scientific projects. We describe plans to evolve PanDA towards a general workload management system for the new BigData initiative announced by the US government. Other planned future improvements to PanDA will also be described