AN INTRODUCTION TO THE OPERATIONAL ANALYSIS OF QUEUING NETWORK MODELS Peter J. Denning, Jeffrey P. Buzen, The Operational Analysis of Queueing Network.

Slides:



Advertisements
Similar presentations
Introduction to Queuing Theory
Advertisements

10. 5: Model Solution Model Interpretation 10
10/12/1999(c) Ian Davis1 Predicting performance Topics: –Operational analysis of network models –Markov models.
Silberschatz, Galvin and Gagne  2002 Modified for CSCI 399, Royden, Operating System Concepts Operating Systems Lecture 19 Scheduling IV.
NETE4631:Capacity Planning (3)- Private Cloud Lecture 11 Suronapee Phoomvuthisarn, Ph.D. /
Queuing Theory For Dummies
Queuing Theory The system is modeled by a stationary stochastic process Jobs are stochastically independent Job steps from device to device follow a Markov.
1 Part II Web Performance Modeling: basic concepts © 1998 Menascé & Almeida. All Rights Reserved.
RAIDs Performance Prediction based on Fuzzy Queue Theory Carlos Campos Bracho ECE 510 Project Prof. Dr. Duncan Elliot.
OS Fall ’ 02 Performance Evaluation Operating Systems Fall 2002.
Single queue modeling. Basic definitions for performance predictions The performance of a system that gives services could be seen from two different.
Data Communication and Networks Lecture 13 Performance December 9, 2004 Joseph Conron Computer Science Department New York University
6/25/2015Page 1 Process Scheduling B.Ramamurthy. 6/25/2015Page 2 Introduction An important aspect of multiprogramming is scheduling. The resources that.
Little’s Theorem Examples Courtesy of: Dr. Abdul Waheed (previous instructor at COE)
1 Multiple class queueing networks Mean Value Analysis - Open queueing networks - Closed queueing networks.
OS Fall ’ 02 Performance Evaluation Operating Systems Fall 2002.
Operational Laws A large number of day-to-day problem in computer systems performance analysis can be solved by using some simple relationships that do.
1 Part VI System-level Performance Models for the Web © 1998 Menascé & Almeida. All Rights Reserved.
Lecture 14 – Queuing Systems
Computer Networks Performance Evaluation. Chapter 12 Single Class MVA Performance by Design: Computer Capacity Planning by Example Daniel A. Menascé,
yahoo.com SUT-System Level Performance Models yahoo.com SUT-System Level Performance Models8-1 chapter12 Single Class MVA.
Introduction to Queuing Theory
OPSM 301: Operations Management
M EAN -V ALUE A NALYSIS Manijeh Keshtgary O VERVIEW Analysis of Open Queueing Networks Mean-Value Analysis 2.
Copyright ©: Nahrstedt, Angrave, Abdelzaher, Caccamo1 Queueing Systems.
Lecture 2 Process Concepts, Performance Measures and Evaluation Techniques.
Performance Evaluation of Computer Systems Introduction
1 Performance Evaluation of Computer Systems and Networks Introduction, Outlines, Class Policy Instructor: A. Ghasemi Many thanks to Dr. Behzad Akbari.
CPE 619 Queueing Networks Aleksandar Milenković The LaCASA Laboratory Electrical and Computer Engineering Department The University of Alabama in Huntsville.
Introduction to Operations Research
NETE4631:Capacity Planning (2)- Lecture 10 Suronapee Phoomvuthisarn, Ph.D. /
Chapter 3 System Performance and Models. 2 Systems and Models The concept of modeling in the study of the dynamic behavior of simple system is be able.
Introduction to Queueing Theory
Queuing Theory Basic properties, Markovian models, Networks of queues, General service time distributions, Finite source models, Multiserver queues Chapter.
CPU Scheduling Gursharan Singh Tatla 1-Feb-20111www.eazynotes.com.
1 Chapters 8 Overview of Queuing Analysis. Chapter 8 Overview of Queuing Analysis 2 Projected vs. Actual Response Time.
Queueing Models with Multiple Classes CSCI 8710 Tuesday, November 28th Kraemer.
OPERATING SYSTEMS CS 3530 Summer 2014 Systems with Multi-programming Chapter 4.
Chapter 10 Verification and Validation of Simulation Models
Chapter 3 System Performance and Models Introduction A system is the part of the real world under study. Composed of a set of entities interacting.
1 Components performance modelling - Outline of queue networks - Mean Value Analisys (MVA) for open and close queue networks.
NETE4631: Network Information System Capacity Planning (2) Suronapee Phoomvuthisarn, Ph.D. /
OPERATING SYSTEMS CS 3530 Summer 2014 Systems and Models Chapter 03.
6.1 CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms Multiple-Processor Scheduling Real-Time Scheduling Algorithm Evaluation.
Ó 1998 Menascé & Almeida. All Rights Reserved.1 Part VI System-level Performance Models for the Web (Book, Chapter 8)
Review Session Jehan-François Pâris. Agenda Statistical Analysis of Outputs Operational Analysis Case Studies Linear Regression.
Internet Applications: Performance Metrics and performance-related concepts E0397 – Lecture 2 10/8/2010.
Little’s Law & Operational Laws. Little’s Law Proportionality relation between the average number of jobs (E[N]) in a system and the average system time.
Copyright ©: Nahrstedt, Angrave, Abdelzaher, Caccamo1 Queueing Systems.
Ó 1998 Menascé & Almeida. All Rights Reserved.1 Part II System Performance Modeling: basic concepts, operational analysis (book, chap. 3)
Queuing Theory Simulation & Modeling.
Ó 1998 Menascé & Almeida. All Rights Reserved.1 Part VI System-level Performance Models for the Web.
OPERATING SYSTEMS CS 3502 Fall 2017
CPE 619 Mean-Value Analysis
Chapter 10 Verification and Validation of Simulation Models
Process Scheduling B.Ramamurthy 11/18/2018.
CPU Scheduling G.Anuradha
Process Scheduling B.Ramamurthy 12/5/2018.
Lecture 2 Part 3 CPU Scheduling
Process Scheduling B.Ramamurthy 2/23/2019.
Process Scheduling B.Ramamurthy 2/23/2019.
Process Scheduling B.Ramamurthy 2/23/2019.
Process Scheduling B.Ramamurthy 4/11/2019.
Process Scheduling B.Ramamurthy 4/7/2019.
Uniprocessor scheduling
Process Scheduling B.Ramamurthy 4/19/2019.
Process Scheduling B.Ramamurthy 4/24/2019.
CSE 550 Computer Network Design
CS 704 Advanced Computer Architecture
Process Scheduling B.Ramamurthy 5/7/2019.
Presentation transcript:

AN INTRODUCTION TO THE OPERATIONAL ANALYSIS OF QUEUING NETWORK MODELS Peter J. Denning, Jeffrey P. Buzen, The Operational Analysis of Queueing Network Models. ACM Computing Surveys 10(3): , 1978.

Queuing Models Computer systems contain many queues Ready queue I/O device queues Message queues … Bottlenecks and queuing delays have a major impact on computer system performance Queuing models capture this impact

Operational Analysis Provides good bounds of performance indices of many computer systems Only makes verifiable assumptions of behavior of these systems Is easier to teach and to understand than stochastic analysis

Applications “Back of the envelope” estimations of system performance Fast and easy Validation of results of a simulation study Checks for reasonableness of results Communication with non-specialists

Other queuing models Stochastic models Also known as Markov models More powerful Often make unrealistic assumptions about investigated systems Disk service times are not exponentially distributed! Provide surprisingly good estimates of computer system performance

Note Stochastic models can also be used to estimate Availability of computer systems: fraction of time system will be operational Reliability of computer systems: probability that the system will never fail over a time interval of duration t Essential for real-time systems and storage systems

BASIC CONCEPTS

Hypothesis testability All hypotheses made by OA can be tested by Observing the behavior of actual systems Over finite time periods

Operational Quantities Can be Basic Quantities: Directly measured over a finite observation period Number of arrivals, … Derived Quantities: Computed from basic quantities Server utilization, mean service time, …

A single server Server has its associated queue We will treat the server and its queue as a single “black box”

Basic Quantities Tthe length of the observation period A the number of arrivals during the observation period Bthe total amount of busy times during the observation period C the number of completions during the observation period

ACB T

Derived quantities = A/Tthe arrival rate X = C/Tthe output rate U = B/Tthe utilization S = B/Cthe mean service time

Example Checkout lane: T = 2 hours A = 31 customers B = 108 minutes C = 30 customers We have = 15.5 customers/hour …

Utilization law U = B/T = (C/T )(B/C) = XS Can compute utilization knowing both Completion rate Mean service time When you think about it, it is fairly intuitive

Job flow balance For most systems, arrivals and completions balance each other over any long observation period Systems have finite buffers We will assume that A = C Can check validity of assumption over any specific observation period If A = C then U = S

Steady state If the system has a steady state, we can define N = average number of tasks in system R = average residency of tasks in system R is also known as the response time

Little’s law If W is the total time spent by all tasks inside the system over the observation period Measured in requests × time units Then N = W/T R = W/C Since W/T = (C/T)(W/C) = XR, N = XR This is an important result

Explanation Number of tasks in the system Time T W N = W/T 0 The green area and the area delimited by the dashed lines have equal sizes

An example An FBI agent has observed people entering and leaving a secret meeting Over a period of two hours he has seen 10 people staying an average of 30 minutes each Having 10 people staying 30 minutes each corresponds to 300 people x minutes This is the same as having 300/120 = 2.5 people staying over the whole duration of the meeting

NETWORKS OF SERVERS

Network of servers (I) Arrivals Departures Open network

Network of servers (II) Arrivals Departures Closed network

Operational quantities Over the observation period, we measure C = the number of job completions C k = the number of tasks completed by device k We define X 0 = C/T = the system throughput X k = C k /T = the output rate at server k V k = C k /C = the visit count at server k

Relations C k = V k C The total number of visits of server k is equal to the number of visits of server k by job completion times the number of job completions X k = V k X 0 The output rate of server k is equal to the number of visits of server k by job completion times the job throughput

Principle of job flow balance For each device i, the output rate X i is the same as the total input rates to device i A i = C i True for all observation periods long enough such that |A i - C i | << C i Output rates X i are then throughputs

System response time (I) We define Nbar = average number of jobs in the system nbar i = average number of jobs at device i Nbar = Σ i nbar i

System response time (II) Applying Little’s law, we have R = Nbar/X 0 and nbar i = R i X i which we can rewrite nbar i = R i V i X 0 Hence R = Σ i V i R i

Application: Bottleneck analysis A system has one CPU and one disk drive It processes transactions such that V CPU = 12 and S CPU = 5ms V Disk = 11 and S DISK = 8ms What is the maximum system throughput?

Bottleneck analysis (cont’d) Let us compute maximum device throughputs Maximum X CPU = 1/0.005 = 200 requests/s Maximum X Disk = 1/0.008 = 125 requests/s Since X i = V i X 0 Maximum throughput compatible with CPU workload is 200/12 = 16.7 transactions/s Maximum throughput compatible with disk workload is 125/11 = 11.4 transactions/s

Bottleneck analysis (cont’d) The disk is this the bottleneck It has highest V i S i product Identifying feature of any bottleneck device Increasing the system throughput might require Sharing disk requests with a second disk Increasing the efficiency of the system I/O buffer Important

Bottleneck analysis (cont’d) In addition, the maximum throughput of 11.4 transactions/s can only be achieved with a 100% disk utilization It would result in very large queues at the disk In practice, we want to limit the disk utilization to 60 to 80%

Systems with terminals M Terminals Whole system

Interactive response time formula We have M terminals Think time Z between the completion of a job and the submission of the next job Applying Little’s law to the whole system M = (R + Z ) X 0 then R = M/X 0 – Z Very Important

Problem 1 A system Can process up to 5 transactions/s Has 60 client workstations Client think time is 5s Can the system achieve a response time of 5 s?

Answer Applying R = M/X 0 – Z, we compute a lower bound for the response time R min = M/X 0,max – Z = 60/5 – 5 =7 Our answer is no

Problem 2 We have M = 50 terminals Z = 20 s R = 4s What is the system throughput?

Answer From R = M/X 0 – Z, we have X 0 = (R + Z)/M Hence X 0 = (20 + 4)/50 = 0.48 tasks/s

Problem 3 Compute the response time of a system knowing the following parameters M = 25 terminals (users) Z = 18 s V disk = 20 visits to the single disk per user interaction U disk = 0.30 S disk = 25 ms ( )

Answer Let us compute first the throughput X 0 Since X k = U k /S k X disk = 0.30/0.025 = 12 disk requests/s Since X k = V k X 0, we have X 0 = X k /V k X 0 = 12/20 = 0.6 interactions/s The response time is then R = M/X 0 – Z = 25/0.6 – 18 = 23.7s

Demand Rather than expressing CPU workloads by the product V CPU S CPU, we can use the total demand of the task D CPU = V CPU S CPU Since X k = U k /S k and X k = V k X 0 U CPU /D CPU = X CPU /V CPU = X 0

Problem 4 A computer system has a single disk It processes tasks with an average CPU demand of 400 ms. Each task requires 60 disk accesses Each disk access takes 10 ms If the CPU utilization is 50%, what are The system throughput? The disk utilization?

Answer X 0 = U CPU /D CPU = 0.5/0.4 = 1.25 tasks/s Since U disk = X disk S disk and X disk = V disk X 0 X disk = 1.25 x 60 = 75 disk requests/s U disk = 75 x =.75 or 75% (a rather high disk utilization)

Load–dependent behavior Device service times may depend on number of jobs at device Disk drives optimize disk accesses Number of visits to a device may depend on workload Swap device We did not cover that topic

Resources ts/Op.Analysis.pdf ts/Op.Analysis.pdf Offers a concise summary of the topics that were discussed in class Peter J. Denning, Jeffrey P. Buzen, The Operational Analysis of Queueing Network Models. ACM Computing Surveys 10(3): , 1978.The Operational Analysis of Queueing Network Models Remains the fundamental text on operational analysis