ANNA UNIVERSITY, CHENNAI
REGULATIONS - 2013
M.E. COMPUTER SCIENCE AND ENGINEERING
CP7001 PERFORMANCE EVALUATION OF COMPUTER SYSTEMS
OBJECTIVES:
To understand the mathematical foundations needed for performance evaluation of computer
systems
To understand the metrics used for performance evaluation
To understand the analytical modeling of computer systems
To enable the students to develop new queueing analysis for both simple and complex systems
To appreciate the use of smart scheduling and introduce the students to analytical techniques for evaluating scheduling policies
UNIT I OVERVIEW OF PERFORMANCE EVALUATION
Need for Performance Evaluation in Computer Systems – Overview of Performance Evaluation
Methods – Introduction to Queueing – Probability Review – Generating Random Variables for
Simulation – Sample Paths, Convergence and Averages – Little’s Law and other Operational Laws – Modification for Closed Systems.
UNIT II MARKOV CHAINS AND SIMPLE QUEUES
Discrete-Time Markov Chains – Ergodicity Theory – Real World Examples – Google, Aloha – Transition to Continuous-Time Markov Chain – M/M/1 and PASTA.
UNIT III MULTI-SERVER AND MULTI-QUEUE SYSTEMS
Server Farms: M/M/k and M/M/k/k – Capacity Provisioning for Server Farms – Time Reversibility and Burke’s Theorem – Networks of Queues and Jackson Product Form – Classed and Closed Networks of Queues.
UNIT IV REAL-WORLD WORKLOADS
Case Study of Real-world Workloads – Phase-Type Distributions and Matrix-Alalytic Methods – Networks with Time-Sharing Servers – M/G/1 Queue and the Inspection Paradox – Task
Assignment Policies for Server Farms.
UNIT V SMART SCHEDULING IN THE M/G/1
Performance Metrics – Scheduling Non-Preemptive and Preemptive Non-Size-Based Policies - . Scheduling Non-Preemptive and Preemptive Size-Based Policies – Scheduling - SRPT and
Fairness.
TOTAL: 45 PERIODS
OUTCOMES:
Upon completion of the course,
the students will be able to
Identify the need for performance evaluation and the metrics used for it Discuss open and closed queueing networks
Define Little’e law and other operational laws
Apply the operational laws to open and closed systems
Use discrete-time and continuous-time Markov chains to model real world systems
Develop analytical techniques for evaluating scheduling policies
REFERENCES:
1. Mor Harchol - Balter, “Performance Modeling and Design of Computer Systems – Queueing
Theory in Action”, Cambridge University Press, 2013.
2. Raj Jain, “The Art of Computer Systems Performance Analysis: Techniques for Experimental
Design, Measurement, Simulation and Modeling”, Wiley-Interscience, 1991.
3. Lieven Eeckhout, “Computer Architecture Performance Evaluation Methods”, Morgan and
Claypool Publishers, 2010.
4. Paul J. Fortier and Howard E. Michel, “Computer Systems Performance Evaluation and
Prediction”, Elsevier, 2003.26
5. David J. Lilja, “Measuring Computer Performance: A Practitioner’s Guide”, Cambridge
University Press, 2000.
6. Krishna Kant, “Introduction to Computer System Performance Evaluation”, McGraw-Hill, 1992.
7. K. S. Trivedi, “Probability and Statistics with Reliability, Queueing and Computer Science
Applications”, John Wiley and Sons, 2001.
No comments:
Post a Comment