Performance evaluation of TCP congestion control algorithms in data center networks

Abstract

TCP congestion control has been known for its crucial role in stabilizing the Internet and preventing congestion collapses. However, with the rapid advancement in networking tech-nologies, resulting in the emergence of challenging network environments such as data center networks (DCNs), the tra-ditional TCP algorithm leads to several impairments. The shortcomings of TCP when deployed in DCNs have moti-vated the development of multiple new variants, including DCTCP, ICTCP, IA-TCP, and D2TCP, but all of these al-gorithms exhibit their advantages at the cost of a number of drawbacks in the Global Internet. Motivated by the belief that new innovations need to be established on top of a solid foundation with a thorough understanding of the existing, well-established algorithms, we have been working towards a comprehensive analysis of various conventional TCP al-gorithms in DCNs and other modern networks. This pa-per presents our first milestone towards the completion of our comparative study in which we present the results ob-tained by simulating multiple TCP variants: NewReno, Ve-gas, HighSpeed, Scalable, Westwood+, BIC, CUBIC, and YeAH using a fat tree architecture. Each protocol is evalu-ated in terms of queue length, number of dropped packets, average packet delay, and aggregate bandwidth as a percent-age of the channel bandwidth.