基于两阶段动态目标的云工作流调度  

Cloud Workflow Scheduling Based on Two-phase Dynamic Objective

作　　者：宋三华[1] SONG Sanhua(College of Information Engineering, Huanghuai University, Zhumadian 463000, Henan, Chin)

机构地区：[1]黄淮学院信息工程学院,河南驻马店463000

出　　处：《实验室研究与探索》2018年第4期134-139,共6页Research and Exploration In Laboratory

基　　金：河南省科技厅科学技术研究基金项目(152102110039)

摘　　要：为了解决云环境中截止时间约束下工作流调度代价优化问题,提出一种基于两阶段动态目标的工作流调度算法TDO-PSO。算法以粒子群进化为基础,定义了工作流任务与资源间的编码机制,设计了满足工作流目标优化的适应度函数。同时,为了适应紧密截止时间约束时可行解搜索困难的问题,设计了两阶段动态目标的搜索模式。在第一阶段,当无法得到可行解时,将满足截止时间约束的最小化执行时间设置为优化目标;在第二阶段,如果获得了可行解,则设置满足截止时间约束的最小化执行代价为优化目标。实验结果表明,TDO-PSO算法不仅可以得到更小的执行代价,且更能够适应紧密截止时间约束。For solving the cost optimization problem of workflow scheduling under deadline constraint in cloud environment,a workflow scheduling algorithm TDO-PSO based on two-phase and dynamic objectives is proposed in this paper. Based on the genetic evolution,we define the code mechanism between workflow tasks and resources,and design the fitness function to meet workflow objective optimization. At the same time,in order to meet the situation of difficultly searching a feasible solution under tight deadline constraint,a two-stage dynamic objectives search pattern is designed.In the first phase,when the genetic feasible solution hasn ’t been obtained,the algorithm focuses on optimizing the execution time objective to meet the deadline constraint. In the second phase,after obtaining a feasible solution,the algorithm focuses on optimizing the execution cost within the deadline constraint. Experimental results show that,TDOPSO can not only find better solution with smaller cost,but is more adaptive to tight deadline constraint situation.

关 键 词：云计算 工作流调度 动态目标 粒子群算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]