Abstract
TCP congestion control algorithm takes an important role in communication throughput. The CUBIC TCP [5] has been widely used in the last decade, especially in server computers. In 2016, the TCP BBR [1] was proposed and some reports showed that the TCP BBR could provide better performance than the CUBIC TCP. In this paper, we evaluate the throughputs of the CUBIC TCP and TCP BBR in conditions wherein the CUBIC TCP and TCP BBR work competitively. We then show a performance cycle with 10 seconds period and reveal that the reason for the cycle is the length of the period of monitoring the minimum round-trip time (RTT) of TCP BBR.