ANALYTICAL MODELING OF A LARGE LOCAL AREA NETWORK - PART I: INTERNET TRAFFIC CHARACTERIZATION

Abstract

The goal of both IP network operators and the end users is to get the highest performance from the system for
a given cost. This makes Performance a key criterion in the design, procurement, and use of computer and
communication systems.
In order to address problems associated with performance degradation of operational communications
systems, over the last decade, traffic engineering techniques have emerged in an attempt to optimize
communication systems’ performance and ensure more efficient use of their resources. One of these
techniques is analytical modeling.
Analytic performance models are an excellent tool for quickly evaluating the performance of operational or
new systems. They are also well suited to comparing the performance of several alternative designs.
However, analytical models can only be developed once detailed knowledge of characteristics of traffic
carried by a network is available. In Part I of this paper, traffic characterization of traffic carried by the
largest Local Area Network (LAN) in Tanzania, University of Dar es Salaam Network (UDSMNET) is carried
out. In Part II of this paper, an analytical model based on the Discrete Time Markov Modulated Poisson
Process is proposed and validated for performance analysis of IP networks.