A low-latency multipath routing without elephant flow detection for data centers

Suchandra Chakraborty, Chien Chen
{"title":"A low-latency multipath routing without elephant flow detection for data centers","authors":"Suchandra Chakraborty, Chien Chen","doi":"10.1109/HPSR.2016.7525638","DOIUrl":null,"url":null,"abstract":"Multipath routing in data center has been studied intensively in recent years in order to minimize the finish time of data transfer or the latency of message sending. Existing scheme of splitting traffic at the granularity of packets can cause reordered packets within a TCP flow. Another alternative scheme of splitting traffic at the granularity of a flow avoids packet reordering but may increase latency as it does not differentiate between elephant and mice flows. A newer efficient approach is to split elephants down into many mice flows, and distribute them uniformly at random over all paths. However, elephant detection is mandatory and the performance is largely depends on accuracy of the elephant detection. We propose, a simple yet effective multipath routing that leverage the same idea of splitting elephants down into many mice flows but without any farther requirement of elephant flow detection by using the hard time-out based flow removal feature of OpenFlow switches. Furthermore, we integrate a VLAN based routing scheme to addresses the challenges of cost-effective scalability. As per our simulation using mininet, we found a 44% reduction in overall flow completion time (FCT) compared to TinyFlow and 32% reduction in the consumption of flow table entry resources inside of aggregate switches in datacenter.","PeriodicalId":342055,"journal":{"name":"2016 IEEE 17th International Conference on High Performance Switching and Routing (HPSR)","volume":"180 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2016-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"11","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2016 IEEE 17th International Conference on High Performance Switching and Routing (HPSR)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/HPSR.2016.7525638","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 11

Abstract

Multipath routing in data center has been studied intensively in recent years in order to minimize the finish time of data transfer or the latency of message sending. Existing scheme of splitting traffic at the granularity of packets can cause reordered packets within a TCP flow. Another alternative scheme of splitting traffic at the granularity of a flow avoids packet reordering but may increase latency as it does not differentiate between elephant and mice flows. A newer efficient approach is to split elephants down into many mice flows, and distribute them uniformly at random over all paths. However, elephant detection is mandatory and the performance is largely depends on accuracy of the elephant detection. We propose, a simple yet effective multipath routing that leverage the same idea of splitting elephants down into many mice flows but without any farther requirement of elephant flow detection by using the hard time-out based flow removal feature of OpenFlow switches. Furthermore, we integrate a VLAN based routing scheme to addresses the challenges of cost-effective scalability. As per our simulation using mininet, we found a 44% reduction in overall flow completion time (FCT) compared to TinyFlow and 32% reduction in the consumption of flow table entry resources inside of aggregate switches in datacenter.
数据中心的低延迟多路径路由,无需大象流检测
为了最大限度地减少数据传输的完成时间或消息发送的延迟,数据中心中的多径路由问题近年来得到了广泛的研究。现有的以数据包粒度划分流量的方案可能导致TCP流中的数据包重新排序。在流的粒度上划分流量的另一种替代方案避免了数据包重新排序,但可能会增加延迟,因为它没有区分大象流和老鼠流。一种较新的有效方法是将大象分成许多老鼠流,并将它们均匀随机地分布在所有路径上。然而,大象检测是强制性的,其性能在很大程度上取决于大象检测的准确性。我们提出了一种简单而有效的多路径路由,它利用了将象群分解为许多老鼠流的相同思想,但通过使用OpenFlow交换机基于硬超时的流去除特性,无需进一步要求象群流检测。此外,我们还集成了基于VLAN的路由方案,以解决成本效益可扩展性的挑战。根据我们使用mininet进行的模拟,我们发现与TinyFlow相比,总体流完成时间(FCT)减少了44%,数据中心聚合交换机内部流表入口资源消耗减少了32%。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信
小红书