An Improved Genetic Algorithm for Solving Bag-of-tasks Scheduling Problems with Deadline Constraints on Hybrid Clouds

摘要

As cloud resources are delivered in a pay-as-you-go manner, users are willing to submit and execute Bag-of-task (BoT) applications on clouds. From the perspective of cloud providers, cloud providers have to outsource some tasks to public clouds with additional costs generated, when their private clouds have insufficient resources to process user-submitted BoT applications with the satisfaction of user-specified quality of service (QoS) requirements. The key issue is how to arrange tasks on the hybrid clouds to maximize profit while meeting those QoS requirements. To solve this problem, we propose an effective improved genetic algorithm (IGA) including a novel crossover, which is able to explore global good genes hiding in the population and inherit them to offspring. Experimental results show that our proposed IGA is not only superior to the standard genetic algorithm, but also outperforms the existing best algorithm (i.e. a particle swarm optimization algorithm) for the considered problem.