| Cloud computing has become one of the hottest topics in research area. Along with the commercialization process of cloud computing, cloud platform has shown a great value for software market. As the core problem of cloud computing, resource allocation and task scheduling has been widely concerned by people. At present,typical algorithms for solving these problems include Max-min algorithm, Min-min algorithm, ant colony algorithm, genetic algorithm, etc. Most traditional task scheduling algorithms are based on cloud service provider and resource level, which ignore user factors. However, as a service oriented technology, cloud computing must protect users' quality of service(Qo S), including makespan, execution cost,bandwidth, system availability, etc. Moreover, traditional task scheduling algorithms have a single optimization object, typically time efficiency, whereas cloud computing must take execution cost and resource utilization ratio as system design factors.In this paper, we propose a new resource allocation algorithm JAACO based on ant colony algorithm, which takes resource usage fairness and system balance into account and makes improvements to above defects. This algorithm classifies tasks according to their Qo S characristics and models tasks by human element model.Meanwhile, JAACO algorithm implements fairness constraint and evaluation constraint in resource selection process by Berger justice allocation principle.JAACO algorithm takes task execution cost and system load balancing into consideration, ensuring decent task makespan, user Qo S and system resource utilization ratio at the same time.We design and implement experiments of ours justice aware ant colony algorithm, i.e., JAACO algorithm based on Cloud Sim and verify the effectiveness and efficiency of JAACO. Experiment results show that our justice aware ant colony algorithm can deal with task Qo S, reduce tasks' execution cost, improve system load balancing and justice. |
PDF Full Text Request