Performance evaluation of computer systems
T Technical Teaching
- The objective of the course is to provide students with simple and efficient methods to analyze the performance of a system.
- Although studied methods rely on mathematical analysis (which will be sketched), the focus will be on the understanding of the methods and the situations in which they can be used (which methods should you use, what to expect, etc.).
- The first part of the course will be dedicated to analysis of performance data (from simulations or experiments); the second part will be dedicated to performance modeling.
- Examples of applications will be given in computer networks, computer systems and in other engineering areas.
The course will cover the following topics:
- Basics of probability (refresher)
- Confidence intervals
- Model fitting and estimation
- Statistical test and analysis of variance
- Markov processes
- Bottleneck analysis
- Single queue, bandwidth sharing and applications to network performance
- Queueing networks and applications
- If time permits, an introduction to more advanced techniques may be given (large system asymptotics, large deviations)
Nb hours: 42.00