A cooperative computing platform for drug discovery and design

This work presents a novel peer-to-peer based application scheduling algorithm for a cycle-sharing cluster computing system for drug discovery and design. Using the idle cycles donated by the clusters scattered over the Internet, data-intensive biologic applications in the field of drug discovery can be computed in this computing system. Clusters that donate idle cycles are organized in peer-to-peer overlay networks to eliminate problems caused by traditional client-server model. A novel application-based scheduling algorithm is designed to assign idle cycles to tasks efficiently. This algorithm also ensures the routing path length on the overlay network is close to the real network routing distance. Solid experimental results show the good performance of the proposal system.