摘要： 目前对云存储产生的网络流量缺乏系统研究，在分析云存储业务实现机制基础上，设计云存储业务自动测试工具，提出云存储业务的网络流量估算模型，并实际测量各种网络服务质量(QoS)条件对云存储用户体验的影响。实验数据表明，云存储业务会产生几十Gb/s的数据洪流。不同类型服务器的事务时延存在差异，云存储对丢包率的容忍度为2%，4 Mb/s~6 Mb/s带宽具有较高性价比。降低网络时延、丢包率、提高接入带宽可提升云存储用户体验。基于网络能力感知的云存储服务架构可以确保较好的网络QoS，与网络扩容相比，具有投资小、部署快、能充分利用现网资源的优势。

关键词: 云存储, 网络流量模型, 服务质量, 时延, 抖动, 丢包, 体验质量

Abstract: Aiming at the systematic research of the network traffic generated by cloud storage services, an automatic test tool is designed based on the analysis of cloud storage services. And an estimation model of network flow is put forward. The user experience under various network Quality of Service(QoS) conditions is also measured. Experimental data show that cloud storage service can generate a data deluge of several Gb/s. There is a gap between the overall transaction times of two different kinds of servers. Packet loss rate lower than 2% has a smaller effect on cloud storage services. Bandwidth between 4 Mb/s and 6 Mb/s is more cost efficient. Either reducing network delay time and packet loss rate or increasing access bandwidth can improve the Quality of Experiece(QoE). A cloud storage service framework based on the perception of capabilities of network devices can ensure better network QoS condition which has advantages of smaller investment, easier deployment and full use of existing network resource compared with traditional method of network expansion.

Key words: cloud storage, network traffic model, Quality of Service(QoS), delay, jitter, packet loss rate, Quality of Experiece(QoE)

中图分类号: 

• TP391