CS 284

Computer Systems Analysis

Fall, 2003

Course Preliminaries

Instructor:            Larry Dowdy

Office:                377 Jacobs Hall

Phone:                322-3031

Email:                  larry.dowdy@vanderbilt.edu

Office Hours:        Mondays 4:30-5:30

                            Wednesdays 4:00-5:00

                            Fridays 2:00-3:00

                            or by appointment

                            or Monday evenings

                            or whenever you can find me

 

Course Homepage: http://www.vuse.vanderbilt.edu/~dowdy/courses/cs284/fall03.html

 

Prerequisite: CS 281 (i.e., a good introductory operating systems course)

 

Text: "Capacity Planning and Performance Modeling: From Mainframes to Client-Server Systems" by D. Menasce, V. Almeida, and L. Dowdy, Prentice-Hall, (1994)

 

Grading:    weekly (Wednesdays) quizzes        25%  (best 10 quizzes)

                 homework sets                              25%

                prediction project with a partner      25%  (service learning)

                comprehensive final exam                25%

Disabilities: Vanderbilt is committed to equal opportunity for students with disabilities. If you have a physical or learning disability, you should ask the Opportunity Development Center to assist you in identifying yourself to your instructor as having a disability, so that appropriate accommodations may be provided.  Absent notification, your instructor assumes that you have no disabilities or seek no accommodation.

Emergency Evacuation Plan:   In the event of a fire or other emergency, the occupants of this class should collect personal belongings, walk out of the room (FGH 211), turn right, and right again down the hall.  Go to the exit at the end of the hall, take Stairway 4 located at the NW end of FGH, and exit the building.  The class is to assemble in front of the Community Partnership House (CPH).  Vanderbilt University policy forbids reentry to a building in which an alarm has occurred without authorization by Vanderbilt Security.  If, in consequence of a disability, you anticipate the need for assistance, please discuss that need with the instructor.

Honor Code:  The Honor Code of Vanderbilt University applies to all work done in CS 284.  The Honor Code stipulates that: 

“I pledge on my honor that I have neither given nor received aid on this work.” 

This applies to all quizzes, examinations, and homework.  This applies to giving,  as well as receiving, unauthorized aid. 

·        However, it is recognized that one learns valuable information from one's peers.  Often, something is unclear from a lecture or the texts or someone understands the presented material in one way while someone else understands the material another way.  Sharing this type of information is strongly encouraged.  BUT … getting a classmate to share how to solve an assigned homework problem is not allowed. Comparing answers of homework problems is not allowed.  Stated differently, sharing of information is okay, sharing of ideas is not.  The basic rule to follow is to ask your instructor if you are in doubt. 

·        Any work handed in to be graded must have been written entirely by you.  The only exception is:  when specified by the instructor that working together in teams is allowed.  In this case, all work must have been written entirely by you and your team members.

·        All students are required to report honor code violations to the instructor.  This is not “squealing” --- it is a course requirement.

            Any suspected violations will be handled in accordance with the guidelines set forward by the Honor Council.

Tentative Schedule: 

Fundamentals (2 weeks) - baby probability, random variables, distributions, expectation, transforms, limit theorems, random processes, Markov processes, Poisson/exponential             processes, birth/death processes  [class notes/outside readings]

Capacity Planning (2 weeks) - the CP problem and general issues, overall methodology, workload characterization, workload forecasting, performance prediction  [MAD chapters 1 -- 2]

Performance Modeling (5 weeks) - balance equations, Little's Law, A/B/D/E/F queues, networks of queues, normalization constants, convolution, non-exponentials, single/multi-class models, open/closed models, operational assumptions, operational laws, bottleneck analysis, performance bounds, mean value analysis, decomposition/aggregation  [MAD chapters 3 -- 6]

Advanced Modeling Topics (2 weeks) - case studies, client server models, load dependent models, priority scheduling, memory modeling, disk modeling, channel modeling, overhead modeling   [MAD chapters 7 -- 8]

 Parameterization (3 weeks) - measurement, monitoring, parameter estimation, sensitivity analysis, model calibration, model validation, case studies  [MAD chapters 9 -- 10]

 Special Purpose Applications (1 week) - new applications, software modeling  [MAD chapter 11]

Comprehensive Final Exam

 

Service Learning Prediction Project Milestones: 

·        September 5 – partnerships formed

·        September 26 – non-profit organization and system selection

·        October 17 – project specification document due (i.e., project goals, system specification, parameterization methodology, model construction methodology, prediction methodology, validation procedures)

·        November 7 – progress report due

·        December 10 – final report due