The so-called information society needs an efficient access to the available information which is often heterogeneous and distributed. In order to make information sharing efficient, some technical solutions have been proposed. The concept of distributed database has been introduced in order to organize a collection of multiple and logically bound databases spread across a computer network. The Peer-to-Peer (P2P) infrastructure is an emergent paradigm offering new opportunities for the conception of large scale distributed systems. Recently the P2P data management systems (Peer Data Management System have appeared). They combine the advantages of the P2P systems with those of the distributed databases. In order to contribute to the research on data management in a P2P context, we propose the SenPeer PDMS. SenPeer is based on a super-peer topology organizing the peers into semantic communities according to their topics of interests. To facilitate data exchange in the presence of heterogeneous schemas and multiple data models, we establish semantic mapping discovery and query reformulation processes. The semantic mappings, in combination with the peer schemas are the basis of a semantic overlay, on top on the underlying physical network and which is used for intelligent query routing. Queries are exchanged through a query exchange format. A distributed optimisation process has the responsibility of finding the best execution plan for the query by taking into account the data distribution and the characteristics of the PDMS.We describe an experimental validation through a simulator to illustrate the feasibility and the performance of the proposed techniques
展开▼
