TCP-friendly marking for scalable best-effort services on the Internet

This paper proposes the use of TCP-awaxe network-based packet marking, which in conjunction with differential packet dropping, is a powerful way to improve the performance of buffer management for best-effort services on the Internet. We extend the notion of "TCP-Friendly" packet marking, proposed recently by us [6], and apply it to improve the performance of traditional best-effort services. The TCP-aware use of deterministic packet marking at the network edge allows us to protect selected TCP packets from suffering loss. This significantly reduces the total number of timeouts and avoids the resulting service degradation. In particular, we protect TCP sessions with very small windows, disperse packet loss across a given window, and protect retransmitted packets from encountering loss. We show baseline results which illustrate that the performance gains are considerable (orders of magnitude) when compared to stateful packet dropping algorithms like FRED [15], and even when TCP SACK [18] implementations are employed. The scheme has been implemented in Linux 2.2.10, and all the results are based upon experimental data.