Imperial College London
ISE-2 Surprise 97 Project
Approaches to Performance Evaluation
Written by Wan Ling Li
Performance - the capabilities of a machine
Benchmarking Benchmarking is a systematic approach to search for and implement an optimum configuration of a system. A benchmark allows the analyst to make comparisons of the current system setup to a modified, and in some cases theoretical model. Obviously with out this benchmark, it would be very hard to calculate the effectiveness of a new system.
A benchmark typically involves measuring the values and parameters in a system. Looking at the car manufacturing example, it is possible to record the rate of cars produced, the time taken between the variety of processes within the plant and the costs of the various systems. Obviously the more events we record the more detailed the performance evaluation will be.
Benchmarking includes four basic categories of activities:
- Planning the benchmarking project
- Collecting data and information from the system
- Analyzing the data
- Improving the system as a result of an analysis of the data
The key to successful benchmarking results are careful research and planning. The goals and objectives of the benchmark must be decided upon at an early stage.
Typically in performance evaluation, there are two main methods to analyzing a system. Either simulation can be used or an analytical approach can be taken. In the former option, the system is closely modeled, accurate results can be obtained, but at the expense of computing time.
Whereas in the latter approach, assumptions are made about the system, which leads to a much faster method of examining the system. The cost with the analytical method is that accuracy is compromised at the expense of speed.
The next few chapters will describe these two methods in more detail.
Last Updated 27th May, 1997