nton.slac.vg.fm April 7, 1998]

works . network-level and platform-level parallelism and pipelining . individual threads of control for every server-level resource . monitoring at all levels for problem diagnosis, algorithm analysis, and application adaptation Reliability (otherwise the system is not real) . autonomous monitoring and management for - system regeneration in the event of transient failure, - system reconfiguration in the event of long-term failure, and - adaptability in the event of degradation 5 Imaging and Distributed Computing Group, Security (otherwise the system is not real) . decentralized management of access rights to match the decentralized nature of the distributed systems: PKI certificate based distributed management of policy 6 Imaging and Distributed Computing Group, Outline of What Follows Overall model Prototype systems . WALDO: A digital library for real-time data sources . STAR: Distributed physics/nuclear science data analysis DPSS: A wide-area network distrib