The proposal describes a DHT-based approach to constructing a scalable and distributed information management system that aggregates information about large scale networked systems. Such a system is envisioned to be a key enabler of large-scale distributed applications through monitoring, querying and reacting to changes in the state of the distributed system(s). A prototype will be developed and experiments in collaboration with TACC are expected to enhance grid infrastructure in the near-term with usable technology to improve data replication, grid information systems, and distributed network monitoring. The proposal addresses new scalability, flexibility, autonomy, and robustness concerns raised by the tight integration of the aggregation tree abstraction with the internal DHT structure. By providing the equivalent of a distributed systems control backplane, middleware infrastructure should be greatly enhanced and simplify distributed application development and use.
