Load Models

This chapter deals with the models of IP packet streams generated by various applications. Initially, the characteristics of load models (correlation, variance) and their impact on QoS values are explained using a simple M/G/1 router model. Subsequently, the question of how to derive a distribution type model from load measurements is answered. The steps from a visual creation of a hypothesis to parameter determination with MLE (Maximum Likelihood Estimation) to distribution testing are explained and implemented as Julia programs. Section 4.4 presents the current state of load model standards. The chapter concludes with the topic of load generators. The suggestion for independent further work on this topic is to test the statement that with an increasing number of aggregated Pareto packet streams, the IAT (Inter Arrival Time) of the packets results in a Weibull distribution.