Measuring TCP Congestion Control Behaviour in the Internet

Abstract

The Internet is constantly changing and evolving. In this thesis the behaviour of various aspects of the implementation of TCP underlying the Internet are measured. These include measures of Initial Congestion Window (ICW), type of reaction to loss, Selective Acknowledgment (SACK) support, Explicit Congestion Notification (ECN) support. We develop a new method to measure congestion window reduction due to three duplicate ACK inferred loss. In a previous study 94% of classified servers showed window halving, whereas we found that 50% of classified servers exhibited Binary Increase Congestion control (BIC) or Cubic style behaviour, which is a departure from a Request For Comments (RFC) requirement to reduce the congestion window by at least 50%. ECN is predicted to improve Internet performance, but previous studies have revealed a low support for it 0.5%, and ECN connections failed at a high rate due to middlebox interference 9%; in this thesis we show a steady increase over time of ECN being implemented and supported 7.2%-10.3%. ECN testing of webservers with globally routable IPv6 adderesses showed a higher success rate 21.9%. Analysis of congestion control behaviour such as Tahoe, Reno and New Reno showed New Reno dominating more strongly than before, increasing from 35% to 70% of popular webservers. SACK sending analysis revealed that 45% of popular webservers implement it properly, as compared to 18% in earlier studies. SACK receiving analysis showed higher results to the earlier studies, with success increasing from 64.7% to 81.1%. For both of these SACK studies results for webservers with globally routable IPv6 addresses showed a higher success rate when errors remained low. Analysis of ICW indicates that 75% of popular webservers implement the older ICW regime of an initial congestion window of two or less packets, as compared to 96% in previous studies. The new regime of an ICW of three or four packets depending on segment size was implemented at 20%. We see from these results that RFCs do affect TCP implementation, but change can be slow. However we see that implementation and support for modern TCP features is increasing.