Real-Time Multipath Transmission Protocol (RMTP): A Software Defined Networks (SDN) based Routing Protocol for Data Centric Networks

2019 International Conference on Electrical, Communication, and Computer Engineering (ICECCE)(2019)

引用 3|浏览0
暂无评分
摘要
In last couple of years, Data Center Networks (DCN) has shown to cement itself as a major infrastructure. With the increasing demands of cloud services and applications, it has also presented that the amount of data being generated is increasing at an alarming rate. This increase in data rate brings with itself several problems related to network management, primary of which constitutes of routing elephant flows alongside mice flows and utilizing all the available resources. Conventional DCN approaches are not up to par with the current amount of data being transmitted in a data center network as most DCN use Equal Cost Multipath protocol (ECMP) which though performs well, however is unable to utilize all the resources. This is mainly due to the selection of shortest path based on available links and selecting the best one. This though helping in network transmission, also presents the issue of under-utilization of spare links. Considering this, Software Defined Network (SDN) is a current network paradigm which can be implemented in a DCN network to make the management of a network easier. This paper presents the Real-Time Multipath Transmission Protocol (RMTP) which makes use of SDN for the regulation of multiple paths in DCN. The proposed RMTP would select the best path in a DCN and send majority of elephant flows on that link. The other remaining links will also be utilized in parallel to transmit the remainder of the data. RMTP was evaluated in an SDN emulated environment and compared against existing DCN protocols. The paper shows that RMPT improves the transmission of big data in terms of throughput, packet loss and throughput during link failure by more than 50%.
更多
查看译文
关键词
-Big Data,Stream Control Transmission Protocol,Software Defined Network,Multipath,Data Centric Network
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要