Massive Data Oriented Replication Algorithms for Consistency Maintenance in Data Grids

Based on the Grid Community (GC) for data usage, differentiated replication algorithms are proposed, by which the replica node is selected according to key replica node (SHN) and the other. First, The replicas are passed into SHN; second, if certain nodes in a GC often access the data resource, it or its nearby nodes become an optional replica node based on the storage and bandwidth. A consistency maintenance algorithm is presented to facilitate the coherence function in a differentiated manner: a pessimistic manner or an optimistic manner, and to update according to the context. The simulation results show that the differentiated mechanism can improve grid access performance.