计算机与现代化Issue(4)：118-126,9.DOI:10.3969/j.issn.1006-2475.2026.04.016
微服务架构中服务网格的策略优化
Strategy Optimization for Service Grids in Microservice Architecture
摘要
Abstract
Microservice architectures,as the current favourite in the application development field,are mostly deployed using Kubernetes platform.In large-scale clusters and high-load environments,the default Kubernetes scheduler may not meet the performance requirements and the default load balancing policy of Kubernetes platform is static and local,which is difficult to deal with the unpredictable and global workloads of microservice applications.To address these issues,The poper proposes new cluster scheduler with global dynamic load balancing algorithm.For the scheduling problem,K-partitioning algorithm is designed based on Karger shrinkage algorithm to partition the applications,and the modules are deployed by using traffic-aware heuristic and most-loaded heuristic packing algorithm;for the load balancing,the GDLB algorithm is designed,which is applied in com-bination with the Istio Service Grid and the Sidecar agent.The experiments are carried out on ExoGENI and Google Kubernetes Engine clusters,and are tested with synthetic applications and Bookstore applications.The results show that the prototype sched-uler has 10%higher successful placement rate and 23%higher average co-located traffic ratio than the Kubernetes default sched-uler.In the load test,the Service Grid-based system has shorter response time and lower CPU and memory consumption.This study effectively optimises the performance of microservice deployment and improves resource utilisation and load balancing.关键词
微服务架构/服务安置/服务网格/负载均衡/资源分配Key words
microservice architecture/service placement/service mesh/load balancing/resource allocation分类
信息技术与安全科学引用本文复制引用
杜森,徐明生,周司徒,张春..微服务架构中服务网格的策略优化[J].计算机与现代化,2026,(4):118-126,9.基金项目
企业中台关键技术研究与应用(XM202337160144) （XM202337160144）
