Overview: This course provides an introduction to the tools and techniques needed to construct and analyze performance models of systems such as computer systems and communication networks. The course covers discrete and continuous time Markov chain models, queues in isolation, queueing networks, fluid queues, and their applications to solving real problems. Several in-depth modeling case studies will be drawn from the areas of parallel and distributed systems, and networks. In addition, some aspects of simulation and measurement will be covered. The goal is to teach fundamentals that will have a long half life.
Prerequisites: An undergraduate level course on probability theory. If you have any doubts about your background, complete the problem set that can be found here. The solutions can be found here here.
Instructor: Professor Don Towsley
Department of Computer Science
towsley at cs . umass dot edu
Class meeting time: M/W 9:05 - 10:20. You should also keep the same time free on Fridays. This time will be used to make up classes due to travel. Go here for the current schedule.
Office hours: TBD
Class WWW site: http://www-net.cs.umass.edu/673. We will make extensive use of the class WWW site. You should check the WWW page on a near daily basis for updates
Class email list: Mailing List There is a mailing list associated with this course, cs673_2008 "at" cs . umass dot edu.
- To add yourself to it, send an email message to majordomo@cs.umass.edu. The body of the email message (not the subject) should contain the line: subscribe cs673_2008 The subject of your mail message does not matter. The instructor will approve it.
- To remove yourself from the class list send an email message to majordomo@cs.umass.edu. The body of the email message (not the subject) should contain the line: unsubscribe cs673_2008 Again, the subject of your mail message does not matter.
Teaching Assistant: TBD
Textbook: There is no textbook for the course. However the following textbook is recommended Stochastic Processes by S.W. Ross. This is not a text on performance evaluation per se but does provide the foundation in probability theory needed to do the course. In addition, I have made the following course notes by Philippe Nain available BASIC ELEMENTS OF QUEUEING THEORY: Application to the Modelling of Computer Systems . These notes were developed while he taught an earlier version of this course. Other references can be found here . In addition, reading material pertinent to the course lectures will be posted online whenever possible. In all other cases, material will be placed on reserve.
Topics:
- Markov chains. Poisson process, discrete and continuous time discrete state Markov chains. Stationarity. Applications.
- Queueing theory: single queue. Little's law, M/M/1 queue, M/G/1 queue, mean value analysis, scheduling policies, priority queues.
- Queueing theory: queueing networks. Product form queueing networks, computational solution algorithms.
- Modeling and solving real world problems.
- Stochastic differential equations and fluid models.
- Measurement-based modeling. MLEs, network tomography
- Self similarity. Heavy-tailed distributions, self-similarity, applications to network traffic modeling.