Performance Modelling

This course presents techniques for modelling the performance of computing and communications systems. It covers tools, techniques, and analytical methods to improve the efficiency or productivity of existing or planned computer systems. It addresses the problem of how to design for the best balance of system behaviour, performance, and workload.

Frequency

This subject has been discontinued; no further courses are planned.

Objectives

At the end of the course, students will have an understanding of the techniques used for analysing the performance of computer and communications systems. They will be able to select appropriate techniques and conduct suitable analyses.

Contents

Introduction:

computer and communications performance modelling; notation; Little's formula

Essentials of probability:

discrete and continuous random variables; generating functions; random partitioning; the Central Limit Theorem

Renewal processes:

the Renewal-Reward theorem; the Inspection Paradox; Poisson processes; case study.

Continuous-time Markov Chains:

equilibrium distributions; reversibility; Kolmogorov's theorem; PASTA.

Queues:

the M/M/1 queue; the M/G/1 queue; multiple servers; finite queue limits; Erlang's formula.

Networks of queues:

quasi-reversibility; product form networks, composition of queues and Markovian routing, open and closed Jackson networks; mean value analysis for networks.

Applications:

priority treatment of control data; flow control via sliding windows; flow control via input buffers; integrated services, first-come-first-served, preemptive priority, movable boundary.

Requirements

There are no formal requirements for this course. Some confidence with basic mathematical analysis would be useful.