Attribute-Based Access to Distributed Data over P2P Networks

Peer-to-peer (P2P) networks are distributed data sharing systems with no dedicated and centralized infrastructure. These systems are attractive because they deliver on the Internet’s promise of true decentralization, offering scalability, availability, fault tolerance, robustness, and low barriers to entry. While P2P systems have been used so far mainly for file sharing, their true potential lies as a vast, loosely connected world wide infrastructure for sharing resources, data and information. However, many challenging research problems must be addressed and solved before this vision can materialize. This paper addresses a natural step in the evolution of P2P: going beyond simple file sharing based on exact-name based lookups to data and information sharing where data is accessed based on its attributes or properties. We have identified diverse applications such as network monitoring, astronomy data applications, event-notification systems, and Grid computing that can benefit directly from attribute-based access to distributed data over P2P systems. Based on the application requirements, we propose three new models for both data distribution and data accesses. For each of these models, we propose can and chord like structures for data storage and information retrieval. A novel aspect of our development is that one of the models identified is directly applicable for building contentbased publish/subscribe systems over P2P networks.