摘要
异构环境中的计算资源在性能和容量方面存在显著差异。差异进一步导致多个调度目标之间产生相互冲突,使得调度成功率较低。为此,面向异构环境,提出一种容器调度任务多目标优化方法。上述方法在考虑异构环境容器调度所需能耗、资源利用率、SLA、负载均衡等因素基础上,以提高资源利用率和负载均衡为目标,构建了容器调度多目标优化目标函数,消除计算资源差异的同时,避免单一目标优化导致的资源碎片问题。在满足节点物理资源约束、Pod从属唯一性约束、约束和修复以及负载均衡多个约束条件下,避免多个调度目标冲突。利用深度睡眠优化器求解目标函数,得出容器调度的多目标优化结果。仿真结果表明,所提方法应用后,容器调度成功率高、异构节点资源使用率高以及集群负载均衡指数低,可以满足异构环境容器调度需求。
Computational resources in heterogeneous environments exhibit significant differences in terms of performance and capacity.These differences further lead to conflicts between multiple scheduling objectives,resulting in a low scheduling success rate.To this end,a multi-objective optimization method for container scheduling task is proposed for heterogeneous environments.Based on the consideration of energy consumption,resource utilization,SLA,load balancing and other factors of container scheduling in heterogeneous environments,the method constructs a multi objective optimization objective function of container scheduling with the goal of improving resource utilization and load balancing,which eliminates the differences in computational resources and avoids resource fragmentation caused by single-objective optimization.Multiple scheduling objective conflicts are avoided under multiple constraints of node physical resource constraints,Pod subordination uniqueness constraints,constraint and repair,and load balancing.A deep sleep optimizer is used to solve the objective function and derive the multi-objective optimization results for container scheduling.The simulation results demonstrate that after the proposed method is applied,the container scheduling success rate is high,the heterogeneous node resource utilization rate is high,and the cluster load balancing index is low,which can meet the requirements of container scheduling in heterogeneous environments.
作者
王凌
黎俊杰
张晓亮
闫梅
WANG Ling;LI Jun-jie;ZHANG Xiao-liang;YAN Mei(Information and Telecommunication Branch,State Grid Jiangsu Electric Power Company Limited,Nanjing Jiangsu 210024,China;State Grid Electric Power Research Institute Co.,Ltd,Nanjing Jiangsu 211100,China;State Grid Information&Telecommunication Branch,Beijing 100053,China)
出处
《计算机仿真》
2026年第1期152-156,共5页
Computer Simulation
基金
国家电网有限公司总部管理科技项目(5700-202318307A-1-1-ZN)。
关键词
异构环境
容器调度
多目标优化
深度睡眠优化器
资源利用率
Heterogeneous environment
Container scheduling
Multi-objective optimization
Deep sleep optimizer
Resource utilization
