• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2014, Vol. 36 ›› Issue (04): 620-626.

• 论文 • 上一篇    下一篇

中国大陆CDN的测量方法研究与性能分析

吴金福,田野   

  1. (1.安徽省高性能计算与应用重点实验室,安徽 合肥 230026;
    2.中国科学技术大学计算机科学与技术学院,安徽 合肥 230026)
  • 收稿日期:2012-09-26 修回日期:2013-02-01 出版日期:2014-04-25 发布日期:2014-04-25
  • 基金资助:

    国家自然科学基金资助项目(61202405);安徽省自然科学基金资助项目(11040606Q52);中央高校基本科研业务费专项资金资助(WK0110000024)

Measurement and performance analysis of
CDNs in mainland China          

WU Jinfu,TIAN Ye   

  1. (1.Anhui Key Laboratory on High Performance Computing,Hefei 230026;
    2.School of Computer Science and Technology,University of Science and Technology of China,Hefei 230026,China)
  • Received:2012-09-26 Revised:2013-02-01 Online:2014-04-25 Published:2014-04-25

摘要:

内容分发网络CDN根据网络的实时环境,对用户请求的内容进行合理的分发,可以有效避免网络拥塞,优化用户体验。CDN已经成为网络基本架构中非常重要的部分。针对CDN的测量,对于分析CDN网络性能、优化CDN网络结构进而提升服务质量都有重要的意义。然而,由于缺乏有效的测量方法,以及中国大陆网络基础设施服务的复杂性,针对中国大陆CDN测量的研究还比较欠缺。提出一种基于HTTP代理服务器作为测量源的CDN测量技术,通过对中国大陆的主要CDN服务商进行广泛深入的测量,获取到不同CDN服务商的节点分布,并分析了CDN节点的部署策略、负载均衡以及不同CDN服务商的节点调度策略。

关键词: 内容分发网络, 网络测量, 节点分布, 部署策略, 负载均衡

Abstract:

Content Delivery Networks(CDNs) are arranged for efficient delivery of the user's requests' according to the realtime environment of the network, avoiding network congestion effectively as well as optimizing the enduser experience. CDNs have become a critical part of today's Internet. Measurements of CDNs are required for understanding CDN performance and analyzing its architecture optimization, which aims to improve the quality of service. Due to the complexity of the network infrastructure in mainland China, researches on measurement and performance analysis of CDNs in mainland China are relatively few. We propose a novel CDNs measurement method based on the HTTP proxy that we use as vantage points. After extensively measuring the CDNs in mainland China, we obtain CDNs nodes distribution. Finally, we analyze the deployment strategy, load balancing, and node scheduling policies of different CDNs service providers.

Key words: content delivery network;network measurement;distribution of nodes;deployment strategy;load balancing