Distributed Query Processing and Catalogs for Peer-to-Peer Systems
Vassilis Papadimos, David Maier, Kristin A. Tufte · 2003
Peer-to-peer (P2P) architectures are commonly used for file-sharing applications. The reasons for P2P's popularity in file sharing -- fault tolerance, scalability, and ease of deployment -- also make it a good model for distributed data management. In this paper, we introduce a scalable P2P framework for distributed data management applications using mutant query plans: XML serializations of algebraic query plan graphs that can include verbatim XML data, references to resource locations (URLs), and abstract resource names (URNs). We show how we can build distributed catalogs based on multihierarchic namespaces that can efficiently handle content indexing and query routing. We also discuss how peers can convey the currency and coverage of their data, and how queries can use this information to manage the inherent tradeoffs between answer completeness, timeliness, and latency.