Mutual Influence of Opposite TCP Flows in a Congested Network

Abstract

With the rapid growth of the Internet community, some of the simple and familiar tasks related to the field of data transfer are becoming increasingly complex. A modern worldwide network can offer high-speed channels and many opportunities for IT companies that provide high load through the Internet. This creates a bunch of new problems for software solutions and algorithms in the field of high-speed digital c ommunications. This article observes one of these problems: the mutual influence between two mutually opposite single-threaded TCP flows with the various congestion control algorithms. In this paper, some of the most efficient congestion control algorithms were tested on a real network using channel emulation equipment. The test results presented in the article show that two-way TCP data transfer with modern congestion control algorithms can lead to a significant performance drop.

Keywords

• tcp
• ip
• highspeed
• congestion control algorithm
• internet
• performance
• bidirectional
• two-way
• 10g networks
• bbr
• cubic
• coexistence