{"title":"An Analytic Goodput Model for TCP Incast","authors":"Zhenqian Feng, Bing Bai, Bao-kang Zhao, Jinshu Su","doi":"10.1109/IMIS.2012.21","DOIUrl":null,"url":null,"abstract":"TCP In cast good put collapse, as a practical problem found in cluster-based storage networks and data center networks, attracts increasing attentions recently. This paper proposes a simple good put model of TCP In cast. Compared with previous In cast model, this model has three different properties. First, the TCP slow start process is considered as an essential part rather than a negligible one, since the request sizes are commonly small and most of the transmissions are accomplished in this process. Second, it exploits a simple logic of determining timeout to account for the cumulative effect of increasing packet losses, reflecting almost all the change points of the good put curves. Finally, the correctness of the model is validated through experimental results instead of simulations.","PeriodicalId":290976,"journal":{"name":"2012 Sixth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing","volume":"38 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2012-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2012 Sixth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IMIS.2012.21","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2
Abstract
TCP In cast good put collapse, as a practical problem found in cluster-based storage networks and data center networks, attracts increasing attentions recently. This paper proposes a simple good put model of TCP In cast. Compared with previous In cast model, this model has three different properties. First, the TCP slow start process is considered as an essential part rather than a negligible one, since the request sizes are commonly small and most of the transmissions are accomplished in this process. Second, it exploits a simple logic of determining timeout to account for the cumulative effect of increasing packet losses, reflecting almost all the change points of the good put curves. Finally, the correctness of the model is validated through experimental results instead of simulations.
TCP在cast good put collapse中出现的实际问题,在基于集群的存储网络和数据中心网络中越来越受到关注。本文提出了一个简单的TCP In cast的good put模型。与以往的In cast模型相比,该模型具有三个不同的特性。首先,TCP慢启动过程被认为是一个重要的部分,而不是一个可以忽略的部分,因为请求大小通常很小,并且大多数传输都是在这个过程中完成的。其次,它利用确定超时的简单逻辑来解释数据包丢失增加的累积效应,反映几乎所有的好放曲线的变化点。最后,通过实验而非仿真验证了模型的正确性。