XML Management in DHT Networks

We consider the problem of eciently sharing large volumes of XML data based on distributed hash table overlay networks. Over the last three years, we have built ViP2P (standing for Views in Peer- to-Peer), a platform for the distributed, parallel dissemination of XML data among peers. At the core of ViP2P stand distributed materialized XML views, dened as XML queries, lled in with data published any- where in the network, and exploited to eciently answer queries issued by any network peer. ViP2P is one of the very few fully implemented P2P platforms for XML sharing, deployed on hundreds of peers in a WAN. This paper describes the system architecture and modules, and the engi- neering lessons learned. We show experimental results, showing that our choices, outperf related systems by orders of magnitude in terms of data volumes, network size and data dissemination throughput.