QOS-Enabled Networks

Miguel Barreiros is the Data Center Practice Lead at Juniper Networks responsible for the EMEA region. Previously he was a Senior Solutions Consultant focused on both Data Centers and IP/MPLS networks Since he joined Juniper Networks in 2006 he has been focused in the creation and development of solutions, and involved in projects that span all stages of building and expanding networks, from design and testing through implementation and ongoing maintenance. He began his networking career in 2000, when as a hobby he was network administrator for a British multiplayer gaming website that hosted network servers for various video games. Miguel has a B.Sc. degree in Electronics and Computer Engineering from Instituto Superior Técnico. He holds Juniper Networks Certificate Internet Expert (JNCIE) 193 and is a Juniper Networks Certified Instructor.Peter Lundqvist, Systems Engineer, Arista Networks, SwedenPeter Lundqvist works in system engineering at Arista Networks since May 2014. Prior to this, he was a senior Beta Engineer at Juniper Networks, supporting the field test and rollout of new features and solutions before they are public available. He has been responsible for the testing of several of the newly deployed QOS features.  Before working in the Engineering department, Peter worked as professional services engineer, in network design and implementation of large-scale networks, several of them including QOS features. Peter worked for more than 13 years at Juniper Networks, prior to which he worked in Cisco Systems as Consulting Engineer.

"My long-time friends Miguel Barreiros and Peter Lundqvist have deep experience designing modern QoS strategies, and they share that experience in this book, from modern QoS building blocks to applied case studies. They’ll equip you well for designing the best QoS approach for your own network."—Jeff Doyle "An excellent overview of the fundamentals of QoS tools and their application, "QoS-enabled Networks" can serve both as an introduction and as a reference. Free from vendor-specific implementation details and configuration knobs, the book focuses on core concepts and on how to apply them to real-world scenarios, making this complex topic come into sharp focus"—Ina Minei, Network Architect, Google"This book addresses the real world scenarios faced by many telcos across the globe. Prioritisation and scheduling at the forefront of network design is key to every telco’s utopia of a fully converged, multiservice network. A great resource in the designers tool kit"—Phill Magill, Head of Network Innovation at Talk Talk Group"This is the first book about QOS that I actually enjoyed reading precisely because the authors focused on real-life QoS and not in academic discussions about it."—Per Nihlen, IP Network Manager, NORDUnet"This book provides a new approach of the complex realm which is the QoS . It offers a detailed theoretical explanation of QoS mechanisms but also great case studies of concrete QoS applications. This book allowed me to better grasp complex QoS configurations such as the hierarchical CoS in BNG environment."—David Roy, IP/MPLS NOC engineer - Orange France"This book contains useful scenarios and real world case studies which expertly convert theoretical knowledge in practical application"—Matheu Leonards, Head of Architecture, Finance and Risk, Thomson Reuters"Mercifully, this book is not a dry academic dissertation on QoS. Rather, it offers an accessibly written and useful insight into the concepts and mechanisms of QoS that augment the tool kit of today’s Network Engineer"—Russell Thompson, Network Engineer, Telstra

• About the Authors xForeword xiPreface xiiiAcknowledgments xvAbbreviations xviPart I THE QOS REALM 11 The QOS World 31.1 Operation and Signaling 41.2 Standards and Per]Hop Behavior 51.3 Traffic Characterization 81.4 A Router without QOS 111.5 Conclusion 12References 12Further Reading 122 The QOS Tools 132.1 Classifiers and Classes of Service 132.2 Metering and Coloring—CIR/PIR Model 152.3 The Policer Tool 162.4 The Shaper Function 172.5 Comparing Policing and Shaping 182.6 Queue 192.7 The Scheduler 212.8 The Rewrite Tool 212.9 Example of Combining Tools 232.10 Delay and Jitter Insertion 272.11 Packet Loss 312.12 Conclusion 32Reference 333 Challenges 343.1 Defining the Classes of Service 353.2 Classes of Service and Queues Mapping 373.3 Inherent Delay Factors 403.4 Congestion Points 463.5 Trust Borders 493.6 Granularity Levels 513.7 Control Traffic 533.8 Trust, Granularity, and Control Traffic 543.9 Conclusion 56Further Reading 564 Special Traffic Types and Networks 574.1 Layer 4 Transport Protocols: UDP and TCP 584.1.1 The TCP Session 614.1.2 TCP Congestion Mechanism 644.1.3 TCP Congestion Scenario 654.1.4 TCP and QOS 664.2 Data Center 674.2.1 SAN Traffic 684.2.2 Lossless Ethernet Networks 694.2.3 Virtualization 714.2.4 Software Defined Networks 734.2.5 DC and QOS 744.3 Real]Time Traffic 744.3.1 Control and Data Traffic 754.3.2 Voice over IP 764.3.3 IPTV 784.3.4 QOS and Real]Time Traffic 79Reference 80Further Reading 80Part II TOOLS 815 Classifiers 835.1 Packet QOS Markings 845.2 Inbound Interface Information 855.3 Deep Packet Inspection 875.4 Selecting Classifiers 885.5 The QOS Network Perspective 895.6 MPLS DiffServ]TE 925.7 Mixing Different QOS Realms 945.8 Conclusion 99References 1006 Policing and Shaping 1016.1 Token Buckets 1016.2 Traffic Bursts 1066.3 Dual]Rate Token Buckets 1096.4 Shapers and Leaky Buckets 1106.5 Excess Traffic and Oversubscription 1126.6 Comparing and Applying Policer and Shaper Tools 1136.7 Conclusion 116Reference 1167 Queuing and Scheduling 1177.1 Queuing and Scheduling Concepts 1177.2 Packets and Cellification 1197.3 Different Types of Queuing Disciplines 1217.4 FIFO 1217.5 FQ 1237.6 PQ 1257.7 WFQ 1277.8 WRR 1287.9 DWRR 1317.10 PB]DWRR 1377.11 Conclusions about the Best Queuing Discipline 141Further Reading 1428 Advanced Queuing Topics 1438.1 Hierarchical Scheduling 1438.2 Queue Lengths and Buffer Size 1468.3 Dynamically Sized versus Fixed]Size Queue Buffers 1498.4 RED 1508.5 Using RED with TCP Sessions 1528.6 Differentiating Traffic inside a Queue with WRED 1548.7 Head versus Tail RED 1568.8 Segmented and Interpolated RED Profiles 1588.9 Conclusion 160Reference 161Further Reading 161Part III CASE STUDIES 1639 The VPLS Case Study 1659.1 High]Level Case Study Overview 1669.2 Virtual Private Networks 1679.3 Service Overview 1689.4 Service Technical Implementation 1709.5 Network Internals 1719.6 Classes of Service and Queue Mapping 1729.7 Classification and Trust Borders 1749.8 Admission Control 1759.9 Rewrite Rules 1769.10 Absorbing Traffic Bursts at the Egress 1799.11 Queues and Scheduling at Core]Facing Interfaces 1799.12 Queues and Scheduling at Customer]Facing Interfaces 1829.13 Tracing a Packet through the Network 1839.14 Adding More Services 1869.15 Multicast Traffic 1889.16 Using Bandwidth Reservations 1909.17 Conclusion 191Further Reading 19110 Case Study QOS in the Data Center 19210.1 The New Traffic Model for Modern Data Centers 19210.2 The Industry Consensus about Data Center Design 19610.3 What Causes Congestion in the Data Center? 19910.3.1 Oversubscription versus Microbursts 19910.3.2 TCP Incast Problem 20210.4 Conclusions 205Further Reading 20711 Case Study IP RAN and Mobile Backhaul QOS 20811.1 Evolution from 2G to 4G 20811.2 2G Network Components 20911.3 Traffic on 2G Networks 21111.4 3G Network Components 21111.5 Traffic on 3G Networks 21511.6 LTE Network Components 21611.7 LTE Traffic Types 21911.8 LTE Traffic Classes 22011.9 Conclusion 224References 227Further Reading 22712 Conclusion 228Index 230