基于动态资源选择策略的微服务工作流调度算法  

Microservice workflow scheduling algorithm based on dynamic resource selection strategy

作　　者：刘洋睿 江凌云[1,2] LIU Yang-rui;JIANG Ling-yun(School of Communication and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;Internet of Things Research Institute,Nanjing University of Posts and Telecommunications,Nanjing 210003,China)

机构地区：[1]南京邮电大学通信与信息工程学院,江苏南京210003 [2]南京邮电大学物联网研究院,江苏南京210003

出　　处：《计算机工程与设计》2023年第5期1313-1319,共7页Computer Engineering and Design

基　　金：江苏省重点研发基金项目(BE2020084-4)。

摘　　要：为解决现有云中工作流调度算法在面对大量微服务任务组成的工作流时出现整体调度成本偏高的问题,提出一种基于动态资源选择策略(dynamic resource selection strategy, DRSS)的微服务工作流调度算法——DRSS调度算法。利用任务在工作流中的位置确定任务的子截止期以及调度优先级,采用动态资源选择策略对任务进行调度,获得任务执行的最优资源,在此基础上更新任务状态以及虚拟机实例的资源向量。实验结果表明,该算法在调度成功率与成本方面上较同类算法更优。To solve the problem of high overall scheduling cost of existing cloud-based workflow scheduling algorithms when dea-ling with workflows consisting of a large number of microservice tasks,a microservice workflow scheduling algorithm based on dynamic resource selection policy,DRSS scheduling algorithm,was proposed.The sub-deadlines and scheduling priorities of tasks were determined using their positions in the workflow,and the tasks were scheduled using the dynamic resource selection policy to obtain the optimal resources for task execution,and the resource vectors of task states and virtual machine instances were updated on this basis.Experimental results show that the proposed algorithm outperforms similar algorithms in terms of scheduling success rate and cost.

关 键 词：云环境 工作流 调度算法 微服务 成本优化 动态资源选择策略 虚拟机实例 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]