基于对等模式的虚拟计算环境下的一种任务调度算法①-计算机系统应用.PDF

2013 年 第 22 卷 第 9 期 计 算 机 系 统 应 用 基于对等模式的虚拟计算环境下的一种任务调度算法① 张 千, 石 琎, 梁 鸿, 关新全 ( 中国石油大学(华东) 计算机与通信工程学院, 青岛 266580) 摘 要: 基于对等模式的虚拟计算系统体系结构,提出一种基于加权随机和反馈机制的任务调度算法. 选定的调 度节点首先对资源进行按需筛选和静态量化排序, 继而对权重递减的资源进行随机选择, 然后获取相应动态信 息进行过载过滤及排序得到最终资源, 最后通过反馈机制实现算法对系统负载的自适应性. 实验表明该算法有 效避免了系统瓶颈, 平衡了负载, 具有较好的自适应性. 关键词: 对等模式; 加权随机; 反馈机制; 任务调度; 负载平衡 A Scheduling Algorithm Based on Peer-to-Peer Mode in Virtual Computing ZHANG Qian, SHI Jin, LIANG Hong, GUAN Xin-Quan (College of Computer and Communication Engineering, China University of Petroleum(East China), Qingdao 266580, China) Abstract: Scheduling algorithm was always a hot topic in virtual computing environment.In order to elimanate system bottleneck and balance load dynamicly. A distributed scheduling algorithm based on weighted random and feedback mechanisms was proposed in this paper. At first the chosen distributed scheduling host chose resources by needs and made static quantification,and then sorted them; Secondly the algorithm chose resources from which sorted by weight randomly, And then it acquired corresponding dynamic information to make load filterate and sort the left. At last it achieved the self-adaptivity to system load through feedback mechanisms. The experiment shows that the algorithm has avoided the system bottleneck effectively and has achieved balanced load as well as self-adaptability to it. Key words: peer-to-peer mode; weighted random;feedback mechanisms; task scheduling; load-balance 虚拟计算环境组织方式[1]主要分为集中式和对等 Min_Min 、Max_Min 算法[2]. 由于OLB 算法是随机把 式, 集中式计算环境实现简单, 且调度节点在了解全 一个可用机器分配给一个待执行任务, 因此机器负载 局资源运行状态下, 实施任务调度, 任务分配合理, 均衡效果较好, 但是因为没有考虑资源释放时间而导 但当系统规模较大时, 过于集中的任务调度处理使得 致其资源利用率较低. Min_Min 算法是将全部未分配 调