Abstract
This tutorial paper describes a procedure for obtaining input parameter values and output performance measures for a popular class of queuemg network models. The procedure makes use of current measurement momtors as much as possible. We survey the two basic approaches to monitoring computer systems (event trace and sampling) and the three types of momtors (hardware, software, and hybrid). Also surveyed are measurement tools for the analytical modeling of several current families of computer systems. We discuss m detail examples of model validations and performance predictions to illustrate the measurement procedures and the class of models