Resource Addressable Network: An Adaptive Peer-to-Peer Discovery Substrate for Internet-Scale Service Platforms

The primary enabling component for using a peer-to-peer (P2P) system as a general service platform is an efficient P2P resource discovery mechanism. The design of a P2P resource discovery mechanism differ from other resource discovery mechanisms because it has to handle the volatile memberships in P2P systems. Controlling the management overhead, implementing fully decentralized mechanisms, and designing against node failures are some of the challenges a P2P discovery service has to handle. I introduce a novel P2P discovery mechanism called resource addressable network (RAN) that is designed to address the above issues. The scalability is a primary concern in the design of the RAN and the constituting algorithms are designed to achieve this goal. RAN can discover resources, i.e., the machines comprising the system, based on both their attributes and their locations. It implements number of supplemental schemes to support this two-tier discovery. RAN’s location-based discovery provides a O(1) route complexity based on the observation that locationing on the Internet is not worthwhile beyond a given resolution. This report provides detailed analysis of resource discovery concepts, presents the design of RAN, compares existing algorithms, presents new algorithms motivated from the drawbacks in the existing ones, and provides simulation results to evaluate the performance of RAN and its components.