实时多任务异构云计算平台负载均衡算法

发布时间：2018-03-15 00:08

本文选题：异构云计算平台　切入点：多任务　出处：《中国科学技术大学学报》2016年03期 　论文类型：期刊论文

【摘要】：针对组成云计算平台各节点之间软件环境存在异构性及数据分布不均匀等原因而导致云计算平台在处理大量任务时往往出现节点负载不均衡的问题,提出了解决异构云计算平台负载均衡方法与相关算法.研究首先统计云计算平台提供的各类服务的平均资源消耗,结合任务分配给指定节点后运行时长和资源占用情况,预测评估某一时刻节点上任务剩余资源消耗需求总量既剩余负载总量;各节点按周期反馈实际任务负载情况,及时修正任务负载信息;最后综合考虑节点各项性能,预测各节点负载评估值,并将待分配任务分发给最适合的节点.实验结果表明,该算法具有可行性并在实时多任务异构云计算平台负载均衡方面具有一定优势.
[Abstract]:In order to solve the problem of heterogeneous software environment and uneven data distribution among the nodes in the cloud computing platform, the cloud computing platform often has the problem of load imbalance when dealing with a large number of tasks. A load balancing method and related algorithms for heterogeneous cloud computing platform are proposed. Firstly, the average resource consumption of all kinds of services provided by the cloud computing platform is counted, and the running time and the resource occupation after the task is assigned to the designated node are analyzed. Forecast and evaluate the total amount of residual load on the node at a certain time; feedback the actual task load according to the period of each node, revise the task load information in time; finally, consider the performance of each node synthetically. The load evaluation value of each node is predicted and the task to be assigned is distributed to the most suitable node. Experimental results show that the algorithm is feasible and has some advantages in load balancing of heterogeneous cloud computing platform with real-time multi-task.
【作者单位】：武汉大学计算机学院;解放军军事经济学院基础部;
【基金】：湖北省重大科技创新计划(2013AAA020) 国家科技重大专项(2013ZX07503-001-06)资助
【分类号】：TP393.09

【相似文献】