摘要
在云服务系统中,流量控制对于保证服务质量非常重要,提出了一种动态两级流量控制策略。利用基于实际云服务系统的改进最小连接数(least connection)算法,计算服务节点的综合负载,对各个服务节点进行负载均衡,并合理分配负载。在服务节点内部,提出Zookeeper动态分级调整阈值的策略,对流量进行动态控制。实际云服务系统性能测试表明,采用动态两级流量控制策略,服务节点的RPS明显优于未采用该策略的RPS,且服务节点的资源利用率有所提高。
Flow control is very important to ensure quality of service in cloud service system. A dynamic two-level flow control strategy was proposed. Using the improved minimum connection number algorithm based on the actual cloud service system, the integrated load of the service node was calculated, load balancing was performed on each service node and the load was distributed reasonably. Within the service node, a dynamic hierarchical threshold adjustment strategy was proposed to control traffic dynamically. The actual cloud service system performance test shows that with the dynamic two-level flow control strategy, the RPS of the service node is significantly better than the RPS without the strategy, and the resource utilization of the service node is improved.
作者
凌其能
王一川
杨贾冰
郑侃
LING Qineng;WANG Yichuan;YANG Jiabing;ZHENG Kan(School of Information and Communication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China)
出处
《物联网学报》
2018年第4期61-67,共7页
Chinese Journal on Internet of Things
基金
国家自然科学基金资助项目(No.61671089)~~
关键词
负载均衡
动态调整
流量控制
云服务系统
load balancing
dynamic adjustment
flow control
cloud service system
