面向多目标的云计算资源调度算法

在传统的虚拟机资源调度中,仅仅考虑当前负载,对虚拟机历史数据没有充分考虑,在处理云计算资源调度的时候出现负载失衡的状况,为了解决上述问题,本文提出了基于启发式遗传算法的资源调度算法,满足多目标规划的情况下实现云计算资源的调度.算法在为用户提供服务的同时充分考虑虚拟机的各种开销和因素,使提供云计算资源的服务器达到负载均衡.对目前的负载情况和历史数据进行分析,经过搜索和计算,计算得到同时满足负载变化数据约束和最小动态迁移开销的最好的云计算资源调度方案.最后,通过仿真实验,对算法进行验证,通过引入负载变化率和平均负载距离二个性能参数来比较和衡量虚拟机负载.实验数据证明,所提出的算法具有很好的全局收敛性和资源利用率,有效解决在资源调度中出现负载失衡和较大动态迁移开销的问题,因此,算法是可行和有效的.

Multi Objective Planning Research of Resource Scheduling Algorithm for Cloud Computing

In the traditional virtual machine scheduling, we only focus on the current load, without fully considering the historical data in the virtual machine. As a result, we will suffer from load imbalance when scheduling the cloud computing resource. In order to solve that problem, this paper puts forward the algorithm of resource scheduling based on heuristic genetic algorithm, which can schedule the cloud computing resource while meeting the multi-objective planning. This algorithm fully considers various overheads and factors of virtual machine while providing service to users, so as to make the server, which provides cloud computing resource, achieve load balancing. By analyzing, researching and calculating current load and historical data, the writer induces the best scheduling scheme of cloud computer resource, which can meet the data constrains for the load variation and minimum dynamic migration overhead. Finally, by verifying the algorithm in a simulation experiment, the writer compares and measures the load of virtual machine by bringing in load change rate and two performance parameters of the average load distance. The experimental data shows that the proposed algorithm has very good global convergence and utilization rate of resources. It can solve the load imbalance and the large overheads of dynamic migration in the process of resource scheduling. Therefore, the algorithm is feasible and effective.