A distributed cross-entropy ANT algorithm for network-aware grid scheduling

Grid scheduling is one of optimally assigning jobs to resources to achieve maximizing the utilization of resources. We propose a distributed ant colony algorithm based on cross-entropy for multi-constraints scheduling. This is an extremely robust rare event simulation technique which may be employed to solve difficult combinatorial optimization problems. We tailor the CE-ANT method for the requirements of network-aware grid scheduling problem. It shows how the task response time can be improved by distinguishing between data-intensive and compute-intensive jobs and scheduling these jobs based on both computational resources and network load. The simulation result demonstrates that the proposed approach succeeds in minimizing the total processing time by at least 10% as compared to its counterpart (Min-Min), and quickly finding the optimal solutions with respect to overhead and speed of convergence compared with ACO. It is highly scalable both in terms of grid site and the number of tasks, indeed it provides superior performance over existing algorithms as the number increase.