A Novel Data Placement Strategy for Science Workflow Based on MCGA in Hybrid Cloud Environment

Abstract

How to accomplish an effective data placement algorithm in hybrid cloud environment has become a crucial issue, especially that science workflow is a sophisticated compute or data-intensive application and brought new challenges by the security issues nowadays. In order to solve the security issues of data placement in hybrid cloud environment, we proposed novel data placement strategy in this paper, and the proposed strategy can be partitioned off three stages. Firstly, the initial feasible solutions are generated stage, it is generated by employing homogeneous method, and we get the initial populations of multilayer coding genetic algorithm; secondly, multi-objects optimize stage trade-off performance and cost, and we balance the resources cost and system performance by using Pareto optimal ideology; finally, generate optimal strategy stage, we estimate Pareto optimal solution and chose the optimum solution act as ultimate data placement strategy. Experiment results prove that our data placement strategy can not only guarantee data security, but also reduce data makespan time.