差异分布式存储系统资源调度的优化仿真  

Optimization Simulation of Resource Scheduling for Differences in Distributed Storage System

作　　者：冯亮[1] 

机构地区：[1]西北工业大学管理学院,陕西西安710072

出　　处：《计算机仿真》2016年第3期343-346,共4页Computer Simulation

摘　　要：针对差异分布式存储系统资源优化调度问题,为提高系统资源利用率,由于分布式存储系统具有一定的活动资源,使得资源特征已发生动态变化,传统的调度算法在进行资源调度时,系统的资源需求量具有强烈的不稳定性,使得资源调度量不能根据之前的调度需求进行分配调度,存在资源调度误差大的问题。提出改进遗传算法的差异分布式存储系统资源调度方法。利用多路径遍历方法获取差异分布式存储系统中全部的资源调度路径集合,选取给定优先权最高的资源任务,建立任务完成时间估算模型,融合遗传算法以任务完成时间最短为优化目标设计适应度函数,计算出的染色体间的交叉变异和适应度的正态分布函数,进行差异分布式存储系统资源调度的优化。仿真结果表明,改进的源调度方法可以大幅度的提高资源调度的效率。A resource scheduling method for differences in distributed storage system is proposed based on im- proved genetic algorithm. The resource scheduling path set is obtained by using muhipath traversal methods, the giv- en resource task with the highest priority is selected, a task completion time estimation model is established, the shor- test time for task completion is taken as the optimization objective to design fitness function via genetic algorithms, the interchromosomal crossover and mutation and the normal distribution of fitness are calculated, the resource scheduling for differences in distributed storage system is optimized. Simulation proves that the proposed source scheduling meth- od can greatly improve the efficiency of resource scheduling.

关 键 词：分布式 资源调度 多路径遍历 

分 类 号：TM744[电气工程—电力系统及自动化]