A twice clustering method was introduced aiming at the case that hybrid cloud environment contains a large number of heterogeneous computing nodes.This method reduced task search space through clustering the heterogeneous resources based on the synthetic characteristics of resources and splitting tasks to the appropriate cluster resources.On this basis, multi-objective task scheduling algorithm in hybrid cloud was proposed, combined with the security and reliability of the private cloud, the scalability of the public cloud and the diversity of user requirements.Firstly, the earliest deadline first algorithm was used in private cloud.To accomplish more tasks, the task was assigned to resource whose completing time was closest to the deadline for each cluster.Then, the public cloud handled the overloading tasks, which would choose the lowest-cost resource under the constraint of computing budget, communication cost and the deadline.Simulation results confirm that the proposed algorithm performs better in lower cost, better resource utilization and greater user satisfaction, compared to the traditional algorithm without clustering.%针对混合云环境包含大量异构云计算节点的情况,提出二次聚类方法,依据资源的综合特性,将异构资源进行分簇,将任务分发到合适的资源聚类,缩小任务搜索空间.在此基础上,结合私有云的安全可靠性、公有云的可扩展性以及用户需求的多样性,提出混合云环境下多目标优化的任务调度算法.该算法首先在私有云优先调度截止时间短的任务,对于每个聚类,将任务分配给完成时间最接近于其结束时间的资源,以完成更多的任务;将溢出的高负载任务转移到公有云聚类执行,结合任务的计算成本、通信开销和截止时间的约束,选择费用最低的资源.实验结果表明,与传统无聚类的算法相比,该算法降低了执行费用,同时提高了资源利用率和用户满意度.
展开▼
