QueueFlower: Orchestrating Microservice Workflows via Dynamic Queue Balancing

doi:10.1109/ICWS62655.2024.00155

	QueueFlower: Orchestrating Microservice Workflows via Dynamic Queue Balancing
	Cao Hongchen; Liu Xinrui; Guo Hengquan; He Jingzhu; Liu Xin
	2024-07-13
会议录名称	2024 IEEE INTERNATIONAL CONFERENCE ON WEB SERVICES (ICWS)
ISSN	2836-3876
页码	1293-1299
发表状态	已发表
DOI	10.1109/ICWS62655.2024.00155
摘要	In microservices, requests’ workflows with the complex dependency graphs pose challenges to auto-scaling strategies. This paper presents QueueFlower, an adaptive and dependency- agnostic auto-scaling framework for orchestrating microservice workflows. QueueFlower leverages real-time latency feedback to estimate queue lengths, effectively identifying congested services without offline profiling. Unlike previous methods that build dependency graphs between services, QueueFlower operates on individual services and adjusts resources proportionally based on estimated queues, ensuring resources of services are balanced globally. We have implemented a prototype of QueueFlower and evaluated its performance on a real-world microservice application. The experimental results demonstrate that compared to baseline methods, QueueFlower significantly reduces request latencies and percentages of SLA violations under stationary and non-stationary workloads.
关键词	Flow graphs Queueing networks Scaling laws Auto-scaling Dependency graphs Microservice Queue lengths Queues balancing Real- time Scalings Servicelevel agreement (SLA) Time latency Work-flows
会议名称	2024 IEEE International Conference on Web Services, ICWS 2024
会议地点	Shenzhen, China
会议日期	7-13 July 2024
URL	查看原文
收录类别	EI
语种	英语
出版者	Institute of Electrical and Electronics Engineers Inc.
EI入藏号	20244817449658
EI主题词	Queueing theory
EISSN	2836-3868
EI分类号	1105 ; 1201.6 ; 1201.8 ; 1301.1 ; 716.1 Information Theory and Signal Processing ; 961 Systems Science
原始文献类型	Conference article (CA)
来源库	IEEE
文献类型	会议论文
条目标识符	https://kms.shanghaitech.edu.cn/handle/2MSLDSTB/424475
专题	信息科学与技术学院_硕士生信息科学与技术学院_博士生信息科学与技术学院_PI研究组_何静竹组信息科学与技术学院_PI研究组_刘鑫组
共同第一作者	He Jingzhu
通讯作者	He Jingzhu; Liu Xin
作者单位	ShanghaiTech University
第一作者单位	上海科技大学
通讯作者单位	上海科技大学
第一作者的第一单位	上海科技大学
推荐引用方式 GB/T 7714	Cao Hongchen,Liu Xinrui,Guo Hengquan,et al. QueueFlower: Orchestrating Microservice Workflows via Dynamic Queue Balancing[C]:Institute of Electrical and Electronics Engineers Inc.,2024:1293-1299.