ECE 577 - Spring 2006
Computer System and Network Evaluation
Syllabus [click here]
Time and Place
Tuesdays and Thursdays, 9:30-10:45am, Harvill Bldg., Room 105
Instructor
Dr. Marwan Krunz
ECE Building, Room 320H
Phone: (520) 621-8731
Email: (krunz@ece.arizona.edu)
Office Hours
Tuesday 11am-12pm, Thursday 4-5pm, and by appointment.
Class Material
There is no required textbook for this class. The material will be based on lecture notes and handouts,
which will either be provided in class or made available through the Copy Center
(Harvill Building, Rm. 137).
References:
- Kishor S. Trivedi, Probability and Statistics with Reliability, Queueing and Computer Science
Applications. John Wiley & Sons Inc., 2002 (2nd edition).
- Raj Jain, The Art of Computer Systems Performance Analysis. John Wiley
& Sons, Inc., 1991.
- L. Kleinrock, Queueing Systems -- Volume I: Theory, J. Wiley & Sons, 1975.
- L. Kleinrock, Queueing Systems -- Volume II: Computer Applications,
J. Wiley & Sons, 1976.
- Research papers.
Prerequistes
ECE 503 or an equivalent course in probability theory and random processes.
Homework Assignments and Solutions
[To be added later]
Course Objectives
Computer systems play a vital role in our lives. The ability to
predict the performance of these systems and optimally design their
parameters is an area of significant interest to computer engineers
and scientists. This course will provide the theoretical
foundation for computer systems analysis and evaluation. With such
foundation, students will learn how to model and evaluate memory systems, CPUs,
network systems, switches, routers, etc. The underlying
principles of computer systems analysis (which are based on
stochastic theory, statistics,
and queueing theory) will be studied. Several operational laws that are
used in analyzing large computer systems will also be discussed.
Topics (tentative):
- Preliminaries: Notation, review of basic concepts in random processes, important theorems,
transform methods, random sums, distribution of failure times, reliability analysis, etc..
- Traffic characterization:
- Elementary traffic models
- Advanced traffic models (Markov models, fluid models, modulated processes, self-similar
processes, etc.).
- Models for multimedia traffic.
- Elementary queueing theory.
- Advanced queueing theory (M/G/1 queue, G/M/1 queue, G/G/1 queue).
- Heavy-traffic approximation.
- Networks of queues: Jackson's networks, open and closed-loop networks.
- Analysis of priority scheduling and queueing systems.
- Fluid analysis.
- Effective bandwidth theory.
- Bounds and approximations.
- Operational laws.
- Workload characterization techniques.
- Mean value analysis (MVA).
- Art of data analysis and representation.
- Statistical techniques: Confidence intervals, analysis of variance, linear and nonlinear regression, etc.
- Analysis of web caching and prefetching systems.
- Other topics (if time permits).
The above topics will be discussed in the context of computer applications (network protocols,
memory systems, capacity analysis, etc.). Examples of related applications will be presented
throughout the course.
Discrete-Event Simulation Using Csim
Although simulations is not the main focus of this course, in some homework assignments you will
be asked to write simulation code and run experiments using the Csim language. The purpose of
these simulations is to study the performance of certain complicated systems that are hard
to analyze or to compare the simulation results with analysis. I will spend 1-2 weeks reviewing
Csim, but that will not be enough to cover all its aspects. Therefore, YOU SHOULD START
LEARNING CSIM ON YOUR OWN AS SOON AS POSSIBLE, AND BEFORE I COVER IT IN CLASS. Csim's User's Guide
is available on electronic reserve and also online from
Mesquite's website (under `Documentation').
Grading:
| Homework | | | 25% |
| Quizzes and Class Participation | | | 10% |
| 1st Midterm Exam (tentatively on Thursday, Feb. 23) | | | 20%
|
| 2nd Midterm Exam (tentatively on Tuesday, April 11) | | | 20%
|
| Final Exam (Tuesday, May 9, 8-10am) | | | 25% |