NATCOR: Stochastic Modelling

Similar documents
M/G/1 and Priority Queueing

Queueing systems. Renato Lo Cigno. Simulation and Performance Evaluation Queueing systems - Renato Lo Cigno 1

Comp 204: Computer Systems and Their Implementation. Lecture 11: Scheduling cont d

Queueing Systems: Lecture 3. Amedeo R. Odoni October 18, Announcements

Computer Networks More general queuing systems

CPSC 531: System Modeling and Simulation. Carey Williamson Department of Computer Science University of Calgary Fall 2017

Contents Preface The Exponential Distribution and the Poisson Process Introduction to Renewal Theory

11 The M/G/1 system with priorities

Section 1.2: A Single Server Queue

Networking = Plumbing. Queueing Analysis: I. Last Lecture. Lecture Outline. Jeremiah Deng. 29 July 2013

Queueing Theory I Summary! Little s Law! Queueing System Notation! Stationary Analysis of Elementary Queueing Systems " M/M/1 " M/M/m " M/M/1/K "

7 Variance Reduction Techniques

2/5/07 CSE 30341: Operating Systems Principles

Introduction to Queueing Theory

Solutions to COMP9334 Week 8 Sample Problems

Introduction to Queueing Theory

Introduction to Queueing Theory

Chapter 6 Queueing Models. Banks, Carson, Nelson & Nicol Discrete-Event System Simulation

Scheduling I. Today Introduction to scheduling Classical algorithms. Next Time Advanced topics on scheduling

CSCE 313 Introduction to Computer Systems. Instructor: Dezhen Song

Part I Stochastic variables and Markov chains

Module 5: CPU Scheduling

Chapter 6: CPU Scheduling

Chapter 2 Queueing Theory and Simulation

CPU scheduling. CPU Scheduling

Queuing Theory. Using the Math. Management Science

Advanced Computer Networks Lecture 3. Models of Queuing

LSN 15 Processor Scheduling

Buzen s algorithm. Cyclic network Extension of Jackson networks

Queueing Review. Christos Alexopoulos and Dave Goldsman 10/6/16. (mostly from BCNN) Georgia Institute of Technology, Atlanta, GA, USA

Queueing Theory and Simulation. Introduction

EE 368. Weeks 3 (Notes)

Slides 9: Queuing Models

Intro to Queueing Theory

Scheduling I. Today. Next Time. ! Introduction to scheduling! Classical algorithms. ! Advanced topics on scheduling

Data analysis and stochastic modeling

Real-time operating systems course. 6 Definitions Non real-time scheduling algorithms Real-time scheduling algorithm

SQF: A slowdown queueing fairness measure

CPU Scheduling. Heechul Yun

Queueing Review. Christos Alexopoulos and Dave Goldsman 10/25/17. (mostly from BCNN) Georgia Institute of Technology, Atlanta, GA, USA

Chapter 1. Introduction. 1.1 Stochastic process

Lecture 20: Reversible Processes and Queues

Elementary queueing system

GI/M/1 and GI/M/m queuing systems

Introduction to Queuing Theory. Mathematical Modelling

An Analysis of the Preemptive Repeat Queueing Discipline

Embedded Systems 14. Overview of embedded systems design

Class 11 Non-Parametric Models of a Service System; GI/GI/1, GI/GI/n: Exact & Approximate Analysis.

Performance Evaluation of Queuing Systems

Link Models for Circuit Switching

Non Markovian Queues (contd.)

CDA5530: Performance Models of Computers and Networks. Chapter 4: Elementary Queuing Theory

Massachusetts Institute of Technology

Fuzzy Queues with Priority Discipline

PBW 654 Applied Statistics - I Urban Operations Research

Online Supplement to Delay-Based Service Differentiation with Many Servers and Time-Varying Arrival Rates

CPU SCHEDULING RONG ZHENG

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Department of Electrical Engineering and Computer Science

A FAST MATRIX-ANALYTIC APPROXIMATION FOR THE TWO CLASS GI/G/1 NON-PREEMPTIVE PRIORITY QUEUE

CHAPTER 4. Networks of queues. 1. Open networks Suppose that we have a network of queues as given in Figure 4.1. Arrivals

Discrete Event and Process Oriented Simulation (2)

CS 550 Operating Systems Spring CPU scheduling I

Monotonicity Properties for Multiserver Queues with Reneging and Finite Waiting Lines

BIRTH DEATH PROCESSES AND QUEUEING SYSTEMS

Control of Fork-Join Networks in Heavy-Traffic

Simulation of Process Scheduling Algorithms

Last class: Today: Threads. CPU Scheduling

NEW FRONTIERS IN APPLIED PROBABILITY

Routing and Staffing in Large-Scale Service Systems: The Case of Homogeneous Impatient Customers and Heterogeneous Servers 1

CS418 Operating Systems

Queueing Theory. VK Room: M Last updated: October 17, 2013.

Andrew Morton University of Waterloo Canada

IOE 202: lectures 11 and 12 outline

Chapter 10. Queuing Systems. D (Queuing Theory) Queuing theory is the branch of operations research concerned with waiting lines.

Basic Queueing Theory

Kendall notation. PASTA theorem Basics of M/M/1 queue

Queueing Theory II. Summary. ! M/M/1 Output process. ! Networks of Queue! Method of Stages. ! General Distributions

Single-Server Service-Station (G/G/1)

Servers. Rong Wu. Department of Computing and Software, McMaster University, Canada. Douglas G. Down

5/15/18. Operations Research: An Introduction Hamdy A. Taha. Copyright 2011, 2007 by Pearson Education, Inc. All rights reserved.

R u t c o r Research R e p o r t. Classes, Priorities and Fairness in Queueing Systems. David Raz a. Benjamin Avi-Itzhak b Hanoch Levy c

Geo (λ)/ Geo (µ) +G/2 Queues with Heterogeneous Servers Operating under FCFS Queue Discipline

Time Reversibility and Burke s Theorem

There are three priority driven approaches that we will look at

Lecture 13. Real-Time Scheduling. Daniel Kästner AbsInt GmbH 2013

Exercises Stochastic Performance Modelling. Hamilton Institute, Summer 2010

Real-time Scheduling of Periodic Tasks (1) Advanced Operating Systems Lecture 2

Process Scheduling. Process Scheduling. CPU and I/O Bursts. CPU - I/O Burst Cycle. Variations in Bursts. Histogram of CPU Burst Times

Queues and Queueing Networks

Embedded Systems Development

4.7 Finite Population Source Model

Operations Research Letters. Instability of FIFO in a simple queueing system with arbitrarily low loads

Job Scheduling and Multiple Access. Emre Telatar, EPFL Sibi Raj (EPFL), David Tse (UC Berkeley)

The M/G/1 FIFO queue with several customer classes

Fair Operation of Multi-Server and Multi-Queue Systems

UC Santa Barbara. Operating Systems. Christopher Kruegel Department of Computer Science UC Santa Barbara

Introduction to Queueing Theory with Applications to Air Transportation Systems

A Fuzzy Approach to Priority Queues

Paper Presentation. Amo Guangmo Tong. University of Taxes at Dallas January 24, 2014

A QUEUE-LENGTH CUTOFF MODEL FOR A PREEMPTIVE TWO-PRIORITY M/M/1 SYSTEM

Transcription:

NATCOR: Stochastic Modelling Queueing Theory II Chris Kirkbride Management Science 2017

Overview of Today s Sessions I Introduction to Queueing Modelling II Multiclass Queueing Models III Queueing Control IV Departure from Exponential Service Times V Time-dependent Behaviour of Queues VI Case Study: Time-dependent Queueing

Overview of this Session Multiclass M/M/1 queue Scheduling Wrap-up

Multiclass M/M/1 queue Arrivals λ 1 λ 2 λ J Queues SCHEDULING Service µ

Scheduling Policies FIFO/FCFS: serve in arrival order LIFO/LCFS: serve in reverse arrival order SPT: shortest processing time first SIRO: serve in random order PS: processor sharing PNPN: priority service, preemptive, non-preemptive Will will consider the first and last policies: M/M/1/ /FIFO M/M/1/ /PNPN

PASTA property PASTA property Poisson Arrivals See Time Averages: In steady-state, the system state seen by an arrival from a Poisson process has the same distribution as that seen by a random observer at a point in time. Intuition: State seen at t is impacted by arrivals before t Interarrival times are exponentially distributed (Memoryless) Interval between t and the next arrival has the same distribution whether t is an arrival instant or not

Utilization Law Utilization Law The probability that a customer of class i is being served is ρ i = λ i /µ i. Let δ i be the probability that class i is being served for i = 1,..., J and Consider W i = W qi + µ 1 i and L i = L qi + δ i. L i = λ i W i and L qi = λ i W qi.

Utilization Law Then L i L qi = λ i W i λ i W qi δ i = λ i /µ i = ρ i. Note: if ρ = ρ 1 + ρ 2 +... + ρ J < 1 then we have steady-state.

Multiclass M/M/1/ /FIFO queue Job from class i arrives. Its service is delayed by: Any ob receiving service W 0 = J ρ µ 1. Jobs awaiting service Hence J L q µ 1 = J ρ W q. W qi = J ρ µ 1 + J ρ W q

Multiclass M/M/1/ /FIFO queue Under FIFO W q1 =... = W qj = W q and so W q = J ρ µ 1 1 J ρ. Application of Little s Law provides W i, L i and L qi for i = 1,..., J.

Multiclass M/M/1/ /PNPN non-preemptive queue Priorities: 1 2... J. Job from class i arrives. Its service is delayed by: Any ob receiving service W 0 = J ρ µ 1. Same or higher priority obs awaiting service i L q µ 1 = i ρ W q. Higher priority obs that arrive while waiting i 1 λ W qi µ 1 i 1 = W qi ρ.

Multiclass M/M/1/ /PNPN non-preemptive queue Hence, i i 1 W qi = W 0 + ρ W q + W qi ρ. The solution of which resolves to the Cobham equations W qi = W 0 (1 i 1 ρ )(1 i ρ ) = J ρ µ 1 (1 i 1 ρ )(1 i ρ ). Other class i performance measures W i, L i and L qi can be obtained via Little s Law.

Multiclass M/M/1/ /PNPN preemptive queue Priorities: 1 2... J. Job from class i arrives. Jobs of lower priority are ignored (preempted by i) Service delayed by same/higher priority obs in service/queue (initial delay) Its service is interrupted by arrivals of higher priority obs (overall service time) Its service is initially delayed by: Ignore obs from class i + 1, i + 2,..., J Invariant of scheduling policy Wait equal to average wait of non-preemptive policy in which classes i + 1, i + 2,..., J don t exist Modified Cobham equations for class i are W I qi = W 0i (1 i 1 ρ )(1 i ρ ) ; W 0i = i ρ µ 1.

Multiclass M/M/1/ /PNPN preemptive queue Average overall service time for a class i ob, τ i, is impacted by preemptions from higher priority obs. Thus W i = W I qi + τ i = τ i = µ 1 = µ 1 i i 1 i + λ τ i µ 1 i 1 + τ i ρ. 1 W 0i (1 i 1 ρ )(1 i ρ ) + µ i 1 i 1 ρ Little s Law gives then provides W qi = W i µ 1 i, L i and L qi.

Wrap up Here, we have: Looked at multiclass queueing models Introduced scheduling policies Queueing models: Extending our models allows us to consider more realistic systems. Question: Is the implementation of a priority rule always of value over a simpler rule (FIFO, say)? Next: Optimal scheduling policies...