Random Models of Performance - Why take a chance?

6/2/99


Click here to start


Table of Contents

Random Models of Performance - Why take a chance?

Introduction

Notions of performance

Mainframe computer

Multi-access systems

Client-server systems

Complex systems

Questions

Some disasters

Capacity planning ...

… fundamentally lacking

Some disasters

New database

Some disasters

Distributed office system

Capacity planning successes

Cache placement

PPT Slide

Work flow management

Breakdown of response time

Location of images

The solution

Types of model

A student’s dilemma

A gambler’s problem

An optimal strategy?

First Result in Performance Modelling

Answer

Single server queue

Queueing Networks

Solution methods

Product-form solution

Solution algorithms

Further developments

Response time distributions

Response times in queueing networks

Open tree-like networks

Closed tree-like networks

Contemporary Methodologies

Queues with negative customers

Response times

Applications

Contemporary Methodologies

Process Algebras

SPA for performance modelling

Some modelling successes

Multi-access computer system

Queueing network model

Predictions

Some modelling successes

ALICE Concrete Architecture

Performance

Performance

Some modelling successes

Client-server Model

Athene model output

Predicted response times

Response time probability density

The future

Performance Engineering

Tools

Close

Author: pgh

Email: pgh@doc.ic.ac.uk

Home Page: http://www.doc.ic.ac.uk/~pgh