摘要
针对算力网络场景下的微服务调度问题,考虑了微服务对网络服务质量的需求和节点能耗优化问题,研究了算力网络场景下服务质量约束的能耗优化微服务调度问题。首先建立了由算力网络调度模块、算力网络调度执行模块和算力网络信息表构成的算力网络调度框架。基于该框架,同时考虑了节点资源限制和网络服务质量约束,建立了能耗优化的微服务调度模型,并采用深度确定性策略梯度算法解决微服务调度问题。最后通过仿真验证了该调度算法的有效性。
Aiming at the micro-service scheduling problem in the computing power network scenario,considering the demand of microservices for network quality of service and the node energy consumption optimization problem,the energy-optimized microservice scheduling problem with service quality constraint in the computing power network scenario is studied.Firstly,it establishes an computing power network scheduling framework consisting of an computing first networking scheduling module,an computing first networking scheduling execution module,and an computing first networking information table.Secondly,based on this framework,considering node resource constraints and network service quality constraints,it establishes an energy-optimized micro-service scheduling model,and solves the micro-service scheduling problem by using a deep deterministic policy gradient algorithm.At last,the effectiveness of the scheduling algorithm is demonstrated by simulation experiments.
作者
刘博文
梁晓晨
张桂玉
韩振东
Liu Bowen;Liang Xiaochen;Zhang Guiyu;Han Zhendong(China Information Technology Designing&Consulting Institute Co.,Ltd.,Beijing 100048,China;China United Network Communications Group Co.,Ltd.,Beijing 100033,China)
出处
《邮电设计技术》
2023年第2期31-36,共6页
Designing Techniques of Posts and Telecommunications
关键词
微服务
能耗优化
服务层协议
算力网络
Micro-services
Energy optimization
Service layer protocols
Computing power network
作者简介
刘博文,助理工程师,硕士,主要从事数据骨干网相关技术研究工作;梁晓晨,工程师,硕士,主要从事数据骨干网相关咨询设计与技术研究工作;张桂玉,毕业于吉林大学,高级工程师,主要从事数据网络的规划、设计工作;韩振东,毕业于清华大学,高级工程师,硕士,主要从事算网基础设施规划编制、新技术研究等相关工作。
