An Approach to Distribute Replicas among Decentralized Tuple Space Nodes

Decentralized tuple space has proven to be a well-suited paradigm for developing distributed component-based systems for its decoupling of processes in space and time, availability and flexibility in system´s organization. However, there are very few research results on data´s replication management and its influence on performance. In particular, we present our approach of tuple replication and various replication policies for specific tuple types. The experiments result shows that applying differentiating replication policies for different scenarios is necessary and achieves performance benefits.