Under the background of big data, cloud computing has attracted plenty of attention. With the continuous development of cloud computing, the application flow such as scientific computing workflow and business workflow, which is restricted by the makespan, energy consumption, cost and other factors, has become more and more complex. Previous cloud application/software can’t meet the needs of businesses and users. Therefore, the cloud workflow system is proposed as a practical and effective solution.Cloud workflow system was capable of making abstract definition for complex workflows, which provided convenience for users. How to deploy tasks of workflow in cloud environment has become a new research object. Task scheduling is the most important core technology of cloud workflow. Cloud computing must pay attention to the quality of service for users and the benefit of the cloud provider due to the characteristics of cloud computing such as user-centric, on-demand service, commercial, heterogeneous environment and so on. Compared to the previous scheduling problem in cloud computing, workflow scheduling must meet not only the constraint of the quality of service (such as time, cost and energy consumption), but also the constraint of the dependence between the tasks of cloud workflow. Furthermore, the intermediate data generated by each task is one of the factors that must be considered.In this paper, we proposed a Multi-objective Particle Swarm Optimization(MOPSO) cloud workflow scheduling strategy according to the characteristics of cloud workflow from the user perspective, which make a compromise between cost and time. On the premise of meeting the QoS requirements for users, the total cost and total time for execution should be reduced. The scheduling policy can return a set that consists of many workflow scheduling schemes for a single workflow instance. This set is a Pareto optimal solution set. Then, an optimal scheduling scheme will be selected based on the user’s preference. In order to further improve the performance of Multi-objective Particle Swarm Optimization, we design a new hybrid algorithm that combine Hill Climbing with Multi-objective Particle Swarm Optimization algorithm(HCMOPSO), the new algorithm will be able to achieve the Pareto optimal solution set with better objective value and can converge more quickly. With the expansion of Workflowsim and the simulation of the workflow applications, the proposed scheduling strategy will be compared with Min-Min, Max-Min and HEFT scheduling algorithms on the extended simulation platform. Experiments show that the new algorithm can get a Pareto optimal solution set in a very short period of time, the solutions outstanding performance in time optimization and significant savings the cost of renting virtual machines and data transmission. |