The Internet is a world wide network of computer networks. Millions of people rely on it to exchange information. Different Internet connections can have different performance reflecting how fast and reliable information can be exchanged between two end-points. Measuring the performance of the connection between two Internet end-points is called end-to-end Internet performance measurement, which plays an important role in the development of the Internet.
End-to-end Internet performance measurement is a measurement technique that injects probe packets into the Internet from a source end-point to a destination end-point. Upon receiving probe packets, the destination end-point decides how to respond to them. If it decides to reply, then some reply packets are sent back to the source. As probe packets travel along a forward path and reply packets travel along a reverse path they experience delay and loss. By measuring these factors we evaluate the performance of the connection between the two end-points.
There are two ways to measure delay and loss: at the source end-point only or at both of the end-points. The former is known as two-way measurement, and examines the performance of the round-trip connection as a whole. The latter is known as one-way measurement, and measures the performance of the connection in the two directions separately.
A popular two-way measurement tool is the ping  utility. Ping sends Internet Control Message Protocol (ICMP; ) request packets to an Internet host and receives ICMP reply packets from that host. The time at which ping sends an ICMP packet and the time at which it receives the reply packet are recorded. The packet Round-Trip Time (RTT) is then calculated by subtracting the time the request packet was sent from the time the reply packet was received.
In this thesis I present a study in which ping packets were sent at high speed to various Internet addresses and their round-trip times were measured by ping and tcpdump , a network monitoring tool. The time ping takes to send ICMP request packets out and get reply packets back in one task is at most several seconds and is considered short.
Two major topics are included in the study: (1) the accuracy of ping and tcpdump, as tools for measuring Round-Trip Times (WRTTs); and (2) patterns of short term high density ping traffic.
This chapter is continued as follows. Section 1.1 states the purposes of the study. Section 1.2 presents the limitations of the study. Section 1.3 defines some important terms used in the thesis. Finally, Section 1.4 outlines the structure of the thesis.