Abstract: To improve Quality of Service (QoS) and reduce operating cost of cloud workflow effectively,the cloud workflow scheduling optimization was researched.Based on analyzing different involving subjects and scheduling links,a cloud workflow scheduling model oriented to QoS and cost-awareness was established.Meanwhile a coding rule dedicated to scheduling scheme was proposed to analyze the scheduling strategy of the proposed model at different stages.A genetic algorithm based on two segment coding of tasks order division was proposed.With tenant leases and virtual machine instance loads as constraints,the population iterative evolution through genetic recombination and mutation processes of two segments was developed,and both cloud service charge saving and cloud resources cost saving were achieved.Simulation experiments were conducted on instances of different sizes,and the results showed that the proposed algorithm could achieve much better solution quality than the two kinds of genetic algorithm based on tasks and virtual machines mapping coding.

Key words: workflow, cloud computing, resources optimization, task scheduling, genetic algorithms

摘要： 为有效提升云工作流服务质量,降低运营成本,对云工作流调度优化问题展开研究。分析问题涉及的不同主体与调度环节,建立面向服务质量与成本感知的云工作流调度模型,并针对问题模型不同阶段的调度策略展开剖析,依据阶段策略特征设计调度方案的编码规则,在此基础上提出一种基于任务序列划分的两段式编码遗传算法。该算法以租户流程租约和虚拟机实例负载为约束,通过两段式交叉、变异算子进行种群的迭代进化,以实现对云工作流服务费用与云资源使用成本的调度优化。通过对不同规模的问题实例进行仿真实验,结果表明所构造算法的解质量明显优于两类基于任务与虚拟机映射编码的遗传算法。

关键词: 工作流, 云计算, 资源优化, 任务调度, 遗传算法