计算机应用研究2025，Vol.42Issue(11)：3405-3411,7.DOI:10.19734/j.issn.1001-3695.2025.04.0122
基于容器镜像共享机制的无服务计算工作流性能优化方法
Performance optimization for serverless computing workflows using container image sharing
摘要
Abstract
To address the challenges of traditional deployment models in serverless computing specifically their difficulty in adapting to dynamic container image resources,which leads to increased cold start latency and low resource utilization,this paper proposed Gust,an optimization approach for serverless computing workflows based on container image sharing.Gust em-ployed a function image similarity-aware scheduling algorithm and a dependency layer restructuring mechanism to enhance the reuse of shared layers across function images,thereby reducing cold start overhead caused by image pulling.It also integrated a genetic algorithm to perform function grouping and partitioning,optimizing workflow deployment and execution across multi-ple nodes.Experiments conducted on the OpenFaaS platform demonstrate the effectiveness of Gust when compared with several baseline methods.Results show that Gust reduces average cold start time by 20.3％and decreases average SLO violation rate by 16.8％.Compared to traditional methods,Gust improves both the execution performance and cost-efficiency of serverless function workflows.关键词
云计算/无服务计算/函数工作流/容器镜像共享机制/部署模型Key words
cloud computing/serverless computing/function workflow/container image sharing mechanism/deployment model分类
计算机与自动化引用本文复制引用
王洋洋,申国伟,崔允贺,付玉杰..基于容器镜像共享机制的无服务计算工作流性能优化方法[J].计算机应用研究,2025,42(11):3405-3411,7.基金项目
国家自然科学基金资助项目(62062022) （62062022）
贵州省自然科学基金重点项目(黔科合基础-ZK[2023]重点011) （黔科合基础-ZK[2023]重点011）
贵州省科技支撑计划项目(黔科合支撑[2023]一般447) （黔科合支撑[2023]一般447）
