CSONG:A Semantic-based Hybrid Overlay Network for Data Grid

The mechanism for a peer to randomly choose logical neighbors, without any knowledge about the physical topology, causes a serious topology mismatch between the overlay networks and the physical network. A flooding-based search is robust to node failure, but the response message will be lost if any node on the response path fails, and the flooding search may causes heavy network traffic. In order to solve these two problems, we propose a semantic-based hybrid overlay network (CSONG). In the tree-like network structure, semantic-based clustering algorithm is used to decide the 1st level super peers and the 2nd level super peers. Node join algorithm, node departure algorithm and query searching algorithm are presented to ensure construction and running of the network. Finally, comparison experiments are made to show the performance of querying search with CSONG.