G-Grid: A Class of Scalable and Self-Organizing Data Structures for Multi-dimensional Querying and Content Routing in P2P Networks

Abstract. Peer-to-Peer (P2P) technologies promise to provide efficient distribution, sharing and management of resources, such as storage, pro-cessing, routing and other sundry service capabilities, over autonomous and heterogeneous peers. Yet, most current P2P systems only support rudimentary query and content routing over a single data attribute, such as the file-sharing applications popularized in Napster, Gnutella and so forth. Full-fledged applications in distributed data management and grid computing demand more complex functionality, including querying and content routing over multiple attributes. In this paper we present a class of scalable and self-organizing multi-dimensional distributed data struc-tures able to efficiently perform range queries in totally decentralized dynamic P2P environments. These structures are not imposed a priori over the network of peers. Rather, they emerge from the independent in-teractions of autonomous peers. They are also adaptive to unanticipated changes in the network topology. This robustness property expands their range of usefulness to many application areas such as mobile ad-hoc net-works.