Sharing of Probabilistically Correlated Data in Peer-To-Peer Networks

Complex query operators usually require joint access§to multiple §data entries whereas single key lookups usually only§affect a single §data entry. The partitioning of the distributed index§of standard §structured overlays is optimized towards single key§lookups and §joint data access as, e.g., required by Peer Data§Management §Systems (PDMS) was neglected so far. (Distributed)§databases have §already shown that (index-)data organization§supporting correlated §data access is necessary and crucial for efficient§processing, as §network usage is minimized. We aim at applying this§insight to §structured overlays by clustering correlated data§frequently accessed §jointly by applications, including PDMS but also§other types of §applications. Data correlations can be derived from§different §sources, data properties, processing properties,§users and §applications. We study and present solutions for§three different §types of correlations in the context of different§applications. Our §approaches are realized on top of the structured§overlay network P-§Grid although they are generic enough to be applied§to other P2P §networks with similar properties.