14.1. Introduction

Efficiency and performance are issues that we have not discussed much so far. We have primarily been concerned with correctness of operation. In this chapter and the next two, we will be focusing not only on the basic tasks TCP performs, but also on how well it performs them. The TCP protocol provides a reliable data delivery service between two applications using an underlying network layer (IP) that may lose, duplicate, or reorder packets. In order to provide an error-free exchange of data, TCP resends data it believes has been lost. To decide what data it needs to resend, TCP depends on a continuous flow of acknowledgments from receiver to sender. When data segments or acknowledgments are lost, TCP initiates a retransmission of the data that has not been acknowledged. TCP has two separate mechanisms for accomplishing retransmission, one based on time and one based on the structure of the acknowledgments. The second approach is usually much more efficient than the first.

TCP sets a timer when it sends data, and if the data is not acknowledged when the timer expires, a timeout or timer-based retransmission of data occurs. The timeout occurs after an interval called the retransmission timeout (RTO). It has another way of initiating a retransmission called fast retransmission or fast retransmit, which usually happens without any delay. Fast retransmit is based on inferring losses by noticing when TCP’s cumulative acknowledgment fails to advance in the ACKs received over time, or when ACKs carrying selective acknowledgment information (SACKs) indicate that out-of-order segments are present at the receiver. Generally speaking, when the sender believes that the receiver might be missing some data, a choice needs to be made between sending new (unsent) data and retransmitting. In this chapter we look closely at how TCP determines that a segment is lost and what to send in response. The issue of how much to send is deferred until Chapter 16, where we discuss TCP’s congestion control procedures that are commonly invoked when packet loss is suspected. Here, we investigate how the RTO is set based on measurements of a connection’s round-trip time (RTT), the mechanics of a timer-based retransmission, and how TCP’s fast retransmission mechanism works. We also look at how SACKs are used to help a TCP sender determine what data is missing at the receiver, the effect of reordering and duplication of IP packets on TCP’s behavior, and the way TCP can change its packet size when retransmitting. We also look briefly at some attacks that can be mounted to fool TCP into behaving more aggressively or more passively.