云计算环境下多目标约束的虚拟资源动态调度

Virtual Resource Dynamic Scheduling with Multi-Objective Constraint in Cloud Computing

为了解决静态资源调度所导致的CPU利用率不高的问题,研究了多目标约束的虚拟资源动态调度方法。给出了云计算虚拟资源调度模型,设计了多目标约束的虚拟资源表示方法,采用马尔科夫链对虚拟资源的下一时刻状态进行预测,从而得到可用资源向量;最后,计算任务与可用资源向量之间的匹配向量,将任务分配给匹配向量中具有最大各维分量之和的虚拟资源进行调度,并提出了具体的采用基于马尔科夫链预测的云计算虚拟资源动态调度算法。实验结果表明:该算法能有效解决云环境下多目标约束的虚拟资源动态调度问题,具有较小的负载均衡离差和任务执行跨度,较其它方法具有较大的优越性。

Aiming at the virtual resource scheduling in cloud computing usually using static scheduling method, only considering the CPU utilization, a virtual resource dynamic scheduling method with multi-goal constraint was proposed. Firstly, virtual resource scheduling model in cloud computing was provided. Then, the representation method with multi goal constraint was designed, the Markov chain was used to predict the state of the virtual resources to obtain the available resources vector, Finally, the calculation match between the tasks and the available resources vector was computed and the task was assigned to resource which has the maximum dimensional components, then the specific dynamic scheduling algorism in cloud computing based on Markov chain was described. The Experiment results show that the algorithm can effectively solve the multi-objective constrain in cloud computing with a smaller load balancing deviation and task makespan, it has larger priority than other methods.