Overview

In modeling computer systems, we typically are interested in the service times of entities that utilize system resources. Entities in our discussion can represent a variety of operations on a computer system. For example, we may be interested in the time it takes to service an operating interrupt or, in a database system, the time to lock a data item in the database. The resources we are interested in are computer hardware elements and software resources. The entities represent the operations that are performed using the resources of the computer system. For example, if the resource is a central processing unit, a program operating on the CPU would have a service time composed of instruction execution (possibly driven by the instruction mix), memory management, I/O management, and secondary device access and transfer delays.

These components of the system under analysis are observable and possibly measurable. This does not mean that we need to measure all components precisely and completely as deterministic points in time. It may actually be more desirable to use average times and random service and arrivals to model these resources and programs. If the focus of review is the overall program operation, and not the components of this operation, then the service times will appear to be unpredictable and, therefore, can be assumed to be random. Without such assumptions, modeling a computer system would get bogged down in the extraction and determination of minute details, which may cloud our overall analysis.

Even though the service times...