摘要
首先介绍了MPTCP协议的技术原理,其次对TCP协议和MPTCP协议的缓存大小和吞吐量的关系进行研究分析,根据研究结果提出了一个针对低延时、高带宽场景的缓存数学计算模型,用于计算MPTCP缓存大小,最后设计了一组本地实验和NorNet Core测试床实验进行分析.实验结果表明,根据模型所设计的缓存取值仅为当前Linux内核缓存默认值的一半,大大减少了资源浪费的同时传输也达到了吞吐量的最大值.
In our report, the technical principle of MPTCP protocol is introduced. The relationship between cache size and throughput of TCP protocol and MPTCP protocol was also analyzed. Based on the research results, a cache calculation model for low latency and high bandwidth scene was established to calculate MPTCP cache size. Lastly, the local experiments and NorNet Core test bed experiment was designed and performed. The results showed that the cache value designed according to the model was only one half of the current Linux kernel cache, which greatly reduced the waste of resources and reached maximum of throughput.
作者
周峰
周星
王琨
廖宇力
符发
Zhou Feng Zhou Xing Wang Kun Liao Yuli Fu Fa(College of Information Science and Technology, Hainan University, Haikou 570228, Chin)
出处
《海南大学学报(自然科学版)》
CAS
2017年第2期111-118,共8页
Natural Science Journal of Hainan University
基金
国家自然科学基金(61363008
61662020)
关键词
多路径传输协议
缓存分析
缓存模型
multipath transport protocol
cache analysis
cache model