Measurement-Based Management of Network Resources

Measurement-Based Management of Network Resources

UCAM-CL-TR-528 Technical Report ISSN 1476-2986 Number 528 Computer Laboratory Measurement-based management of network resources Andrew William Moore April 2002 JJ Thomson Avenue Cambridge CB3 0FD United Kingdom phone +44 1223 763500 http://www.cl.cam.ac.uk/ c 2002 Andrew William Moore This technical report is based on a dissertation submitted by the author for the degree of Doctor of Philosophy to the University of Cambridge. Technical reports published by the University of Cambridge Computer Laboratory are freely available via the Internet: http://www.cl.cam.ac.uk/TechReports/ Series editor: Markus Kuhn ISSN 1476-2986 Summary Measurement-Based Estimators are able to characterise data flows, enabling improvements to existing management techniques and access to previously impossible management techniques. It is the thesis of this dissertation that in addition to making practical adaptive management schemes, measurement-based estimators can be practical within current limitations of resource. Examples of network management include the characterisation of current utilisation for explicit admission control and the configuration of a scheduler to divide link-capacity among competing traffic classes. Without measurements, these management techniques have relied upon the ac- curate characterisation of traffic — without accurate traffic characterisation, network resources may be under or over utilised. Embracing Measurement-Based Estimation in admission control, Measurement-Based Admis- sion Control (MBAC) algorithms have allowed characterisation of new traffic flows while adapt- ing to changing flow requirements. However, there have been many MBAC algorithms proposed, often with no clear differentiation between them. This has motivated the need for a realistic, implementation-based comparison in order to identify an ideal MBAC algorithm. This dissertation reports on an implementation-based comparison of MBAC algorithms con- ducted using a purpose built test environment. The use of an implementation-based comparison has allowed the MBAC algorithms to be tested under realistic conditions of traffic load and re- alistic limitations on memory, computational resources and measurements. Alongside this com- parison is a decomposition of a group of MBAC algorithms, illustrating the relationship among MBAC algorithm components, as well as highlighting common elements among different MBAC algorithms. The MBAC algorithm comparison reveals that, while no single algorithm is ideal, the specific resource demands, such as computation overheads, can dramatically impact on the MBAC algo- rithm’s performance. Further, due to the multiple timescales present in both traffic and manage- ment, the estimator of a robust MBAC algorithm must base its estimate on measurements made over a wide range of timescales. Finally, a reliable estimator must account for the error resulting from random properties of measurements. Further identifying that the estimator components used in MBAC algorithms need not be tied to the admission control problem, one of the estimators (originally constructed as part of an MBAC algorithm) is used to continuously characterise resource requirements for a number of classes of traffic. Continuous characterisation of traffic, whether requiring similar or orthogonal resources, leads to the construction and demonstration of a network switch that is able to provide differentiated service while being adaptive to the demands of each traffic class. The dynamic allocation of resources is an approach unique to a measurement-based technique that would not be possible if resources were based upon static declarations of requirement. 3 4 Acknowledgements I would like to thank my supervisors, Simon Crosby and Ian Leslie, for encouragement and valuable advice during the course of my research. I am grateful to many people for their cajoling, enthusiasm, encouragement and assistance. At the risk of missing-out people whom I should not, I thank current and former members of the Systems Research Group, including Richard Black, Austin Donnelly, Tim Granger, Steven Hand, Rebecca Isaacs, Richard Mortier, Ian Pratt, Kerry Rodden and Neil Stratford. My thanks to my colleagues in the Computer Laboratory, notably Neil Dodgson, Margaret Levitt, Kona McPhee and Lewis Tiffany. My support from college has been greatly appreciated — my thanks in particular to Margaret Cathie, David Greaves, Paul Hewitt and Andy Hopper. My thanks to members of my old department, none of whom ever doubted this would all come to fruition. Thanks in particular to Jim Breen, Eryn Glover, Grant Hampson, Tony McGregor and Henry R. Wu. I am grateful to the researchers who have willingly answered my questions and discussed my proposals, notably Ian Graham, Matt Grossglauser, Richard Gibbens, Suigh Jamin, Frank Kelly, Peter Key, Ed Knightly, Jingyu Qiu, Scott Shenker, Jonathan Smith and Gary Walley. I am indebted to Ian Leslie, Steven Hand and Rebecca Isaacs who have read and commented on earlier drafts of this dissertation, and extend special thanks to Ralphe Neill and Philippa Baines who proof-read the final text. This work was supported by scholarships from the Computer Laboratory, Corpus Christi Col- lege and the Cambridge Commonwealth Trust. My thanks to these orgainisations for making it possible for me to embark upon this work. Finally, I thank my family and friends geographically near and far for their love and support. 5 6 Contents List of Figures 11 List of Tables 13 Glossary 15 1 Introduction 19 1.1 Motivation . 20 1.2 Context . 23 1.3 Contribution . 25 1.4 Outline . 27 2 Background 29 2.1 Network Traffic . 29 2.1.1 Source Modelling . 30 2.1.2 Network Modelling . 37 2.1.3 Fixed-Point and Behavioural Network Modelling . 39 2.2 Traffic Used in this Study . 40 2.2.1 TP10S1 — 2-state ON-OFF Markov model . 41 2.2.2 PP10S1 — 2-state ON-OFF Pareto model . 42 2.2.3 VP64S64 — Voice channel uncompressed . 42 2.2.4 VP64S23 — Voice channel with compression . 42 2.2.5 VP25S4 — Video data stream . 42 2.2.6 RP10S1 — Internet LAN traffic . 44 2.2.7 EP6S480k — Internet WAN traffic . 46 2.2.8 WP10S1 — Elastic WWW traffic . 46 2.3 Network Control . 47 2.3.1 Timescales . 48 2.3.2 Packet Scheduling Level . 50 2.3.3 Burst Level . 53 2.3.4 Session Level . 56 2.3.5 Beyond the Session Level . 61 2.4 Measurement . 62 7 2.5 Summary . 68 3 Environment 71 3.1 Introduction . 71 3.2 Background and Previous Work . 72 3.3 Test environment construction . 74 3.3.1 Network switch . 74 3.3.2 Measurement controller . 77 3.3.3 Traffic Generator . 77 3.3.4 Traffic generator controller . 84 3.3.5 Flow Generator . 85 3.3.6 Admission Controller . 85 3.3.7 Packet time-frame scaling . 86 3.4 Test environment operation . 87 3.5 Test Environment Evaluation . 89 3.5.1 Traffic Generator . 90 3.5.2 Run length and initial stability . 90 3.5.3 Performance . 93 3.5.4 Repeatability . 95 3.6 Summary . 98 4 MBAC algorithms 101 4.1 Introduction . 101 4.1.1 Approach . 102 4.2 Estimators . 104 4.2.1 E-IU — Instantaneous Utilisation . 105 4.2.2 E-CB — Chernoff Bounds . 105 4.2.3 E-MS — Measured Sum . 108 4.2.4 E-MPFE — Measure Per-Flow Estimator . 109 4.2.5 E-MAE — Measure Aggregate Estimator . 112 4.2.6 E-MVE — Mean Variance Estimator . 115 4.2.7 E-KQ — Traffic Envelope . 116 4.2.8 E-GT — Time-scale Decomposition . 118 4.2.9 E-LBE — Loss-Based Estimator . 121 4.2.10 E-GAN — Equivalent Capacity . 123 4.2.11 E-BD — Exponential Upper-Bounds . 124 4.2.12 E-EMW — Effective Bandwidth Model . 127 4.2.13 Estimator Summary . 129 4.3 Policies . 131 4.3.1 P-T — Target . 132 4.3.2 P-TO — Threshold Only . 134 4.3.3 P-BP — Back-off Period . 134 4.3.4 P-PA — Pessimistic Admission . 138 8 4.3.5 P-AR — Policy of AC-AR . 140 4.3.6 Policy Summary . 140 4.4 AC Algorithms . 142 4.4.1 AC-PRA — Peak-rate Allocation . 144 4.4.2 AC-ST — Simple Threshold . 145 4.4.3 AC-AR — Acceptance Region . 145 4.4.4 AC-CB — Chernoff Bounds . 147 4.4.5 AC-MS — Measured Sum . 148 4.4.6 AC-MPFE — Measure Per-Flow Estimator . 149 4.4.7 AC-MAE — Measure Aggregate Estimator . 150 4.4.8 AC-MVE — Mean-Variance Estimator . 150 4.4.9 AC-KQ — Traffic Envelope . 151 4.4.10 AC-GT — Time-scale Decomposition-based Estimator . 153 4.4.11 AC-LBE — Loss-Based Estimator . 153 4.4.12 AC-GAN — Equivalent Capacity . 154 4.4.13 AC-BD — Exponential Upper-Bounds . 154 4.4.14 AC-EMW — Effective Bandwidth Model . 155 4.4.15 AC-T — Target . 155 4.4.16 AC Summary . 156 5 Comparing MBAC algorithms 159 5.1 Introduction . 159 5.1.1 Objectives . 159 5.1.2 Structure . 159 5.2 Method . 160 5.2.1 Criteria . ..

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    273 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us