Art of Network Architecture, The: Business-Driven Design, 1st edition

  • Russ White
  • Denise Donohue


The Art of Network Architecture

Business-Driven Design


The business-centered, business-driven guide to architecting and evolving networks


The Art of Network Architecture is the first book that places business needs and capabilities at the center of the process of architecting and evolving networks. Two leading enterprise network architects help you craft solutions that are fully aligned with business strategy, smoothly accommodate change, and maximize future flexibility.


Russ White and Denise Donohue guide network designers in asking and answering the crucial questions that lead to elegant, high-value solutions. Carefully blending business and technical concerns, they show how to optimize all network interactions involving flow, time, and people.


The authors review important links between business requirements and network design, helping you capture the information you need to design effectively. They introduce today’s most useful models and frameworks, fully addressing modularity, resilience, security, and management. Next, they drill down into network structure and topology, covering virtualization, overlays, modern routing choices, and highly complex network environments.


In the final section, the authors integrate all these ideas to consider four realistic design challenges: user mobility, cloud services, Software Defined Networking (SDN), and today’s radically new data center environments.


•  Understand how your choices of technologies and design paradigms will impact your business

•  Customize designs to improve workflows, support BYOD, and ensure business continuity

•  Use modularity, simplicity, and network management to prepare for rapid change

•  Build resilience by addressing human factors and redundancy

•  Design for security, hardening networks without making them brittle

•  Minimize network management pain, and maximize gain

•  Compare topologies and their tradeoffs

•  Consider the implications of network virtualization, and walk through an MPLS-based L3VPN example

•  Choose routing protocols in the context of business and IT requirements

•  Maximize mobility via ILNP, LISP, Mobile IP, host routing, MANET, and/or DDNS

•  Learn about the challenges of removing and changing services hosted in cloud environments

•  Understand the opportunities and risks presented by SDNs

•  Effectively design data center control planes and topologies


Table of contents

Introduction xx

Part I Framing the Problem

Chapter 1 Business and Technology 1

Business Drives Technology 2

    The Business Environment 2

        The Big Picture 3

        The Competition 4

    The Business Side of the Network 5

        Technologies and Applications 5

        Network Evaluation 6

    The Network’s Customers 6

        Internal Users 7

        External Users 8

        Guest Users 9

Technology Drives Business 9

Part II Business-Driven Design

Chapter 2 Designing for Change 11

Organic Growth and Decline 12

Mergers, Acquisitions, and Divestments 14

Centralizing Versus Decentralizing 15

Chapter 3 Improving Business Operations 19

Workflow 19

    Matching Data Flow and Network Design 20

        Person-to-Person Communication 21

        Person-to-Machine Communication 21

        Machine-to-Machine Communication 22

        Bringing It All Together 23


    BYOD Options 24

    BYOD Design Considerations 27

    BYOD Policy 28

Business Continuity 29

    Business Continuity Versus Disaster Recovery 29

    Business Continuity Planning 30

    Business Continuity Design Considerations 31

Summary 33

Part III Tools of the Trade

Chapter 4 Models 35

The Seven-Layer Model 36

    Problems with the Seven-Layer Model 38

The Four-Layer Model 38

Iterative Layering Model 39

    Connection-Oriented and Connectionless 41

A Hybrid Model 42

    The Control Plane 43

        What Am I Trying to Reach? 43

        Where Is It? 44

        How Do I Get There? 45

        Other Network Metadata 46

    Control Plane Relationships 46

        Routing 46

        Quality of Service 48

        Network Measurement and Management 49

        Interaction Between Control Planes 49

Reactive and Proactive 51

The Waterfall Model 53

Places in the Network 54

Summary 56

Chapter 5 Underlying Support 57

Questions You Should Ask 57

    What Happens When the Link Fails? 57

    What Types of Virtualization Can Be Run Over This Link? 58

    How Does the Link Support Quality of Service? 59

        Marking Packets 59

        Queues and Rate Limiters 59

        Speeds and Feeds Versus Quality of Service 60

Spanning Tree 61


    TRILL Operation 62

    TRILL in the Design Landscape 64

    TRILL and the Fabrics 65

Final Thoughts on the Physical Layer 65

Chapter 6 Principles of Modularity 67

Why Modularize? 68

    Machine Level Information Overload 68

        Machine Level Information Overload Defined 69

        Reducing Machine Information Level Overload 71

        Separating Complexity from Complexity 72

    Human Level Information Overload 73

        Clearly Assigned Functionality 74

        Repeatable Configurations 75

    Mean Time to Repair and Modularization 75

How Do You Modularize? 77

    Topology and Reachability 77

        Aggregating Topology Information at Router B 78

        Aggregating Reachability Information at Router B 78

        Filtering Routing Information at Router B 79

    Splitting Failure Domains Horizontally and Vertically 79

Modularization and Optimization 81

Summary 82

Chapter 7 Applying Modularity 83

What Is Hierarchical Design? 83

    A Hub-and-Spoke Design Pattern 84

    An Architectural Methodology 85

        Assign Each Module One Function 85

        All Modules at a Given Level Should Share Common Functionality 86

        Build Solid Redundancy at the Intermodule Level 87

        Hide Information at Module Edges 88

Typical Hierarchical Design Patterns 89

Virtualization 90

    What Is Virtualization? 90

        Virtualization as Vertical Hierarchy 93

    Why We Virtualize 93

        Communities of Interest 94

        Network Desegmentation 94

        Separation of Failure Domains 94

    Consequences of Network Virtualization 95

Final Thoughts on Applying Modularity 96

Chapter 8 Weathering Storms 97

Redundancy as Resilience 98

    Network Availability Basics 98

    Adding Redundancy 99

MTTR, Resilience, and Redundancy 100

    Limits on Control Plane Convergence 100

    Feedback Loops 102

    The Interaction Between MTTR and Redundancy 103

Fast Convergence Techniques 104

    Detecting the Topology Change 104

    Propagating Information About the Change 105

    Calculating the New Best Path 106

    Switching to the New Best Path 107

    The Impact of Fast Convergence 107

Fast Reroute 108

    P/Q Space 109

    Loop-Free Alternates 110

    Remote Loop-Free Alternates 110

    Not-Via Fast Reroute 111

    Maximally Redundant Trees 113

    Final Thoughts on Fast Reroute 115

The Human Side of Resilience 115

Chapter 9 Securing the Premises 117

The OODA Loop 118

    Observe 119

    Orient 122

    Decide 124

    Act 125

Brittleness 125

Building Defense In 126

    Modularization 128

        Modularity, Failure Domains, and Security 128

        Modularity, Complexity, and Security 128

        Modularity, Functionality, and Security 129

    Resilience 129

Some Practical Considerations 129

    Close a Door, Open a Door 129

    Beware of Virtualization 131

    Social Engineering 131

Summary 132

Chapter 10 Measure Twice 133

Why Manage? 133

    Justifying the Cost of the Network 134

    Planning 135

    Decreasing the Mean Time to Repair 136

    Increasing the Mean Time Between Mistakes 136

Management Models 137

    Fault, Configuration, Accounting, Performance, and Security 137

    Observe, Orient, Decide, and Act (OODA) 138

Deploying Management 140

    Loosen the Connection Between Collection and Management 140

    Sampling Considerations 141

    Where and What 142

        End-to-End/Network 142

        Interface/Transport 143

        Failure Domain/Control Plane 143

Bare Necessities 144

Summary 145

Part IV Choosing Materials

Chapter 11 The Floor Plan 147

Rings 147

    Scaling Characteristics 147

    Resilience Characteristics 149

    Convergence Characteristics 151

    Generalizing Ring Convergence 154

    Final Thoughts on Ring Topologies 155

Full Mesh 155

Clos Networks 157

    Clos and the Control Plane 159

    Clos and Capacity Planning 160

Partial Mesh 161

Disjoint Parallel Planes 162

    Advantages of Disjoint Topologies 163

    Added Complexity 164

    The Bottom Line 164

Divergent Data Planes 165

Cubes 166

Toroid Topologies 167

Summary 169

Chapter 12 Building the Second Floor 171

What Is a Tunnel? 171

    Is MPLS Tunneling? 173

Fundamental Virtualization Questions 175

    Data Plane Interaction 176

    Control Plane Considerations 177

        Control Plane Interaction 177

        Scaling 178

    Multicast 179

    Security in a Virtual Topology 180

MPLS-Based L3VPNs 182

    Operational Overview 182

    Fundamental Questions 185

    The Maximum Transmission Unit 185

    Quality of Service 186

    Control Plane Interaction 186

    Scaling 187

    Multicast 188

    Security in MPLS-Based L3VPNs 188

    MPLS-Based L3VPN Summary 188


    Operational Overview 189

    Fundamental Questions 190

    Control Plane Interaction 190

    Scaling 190

    VXLAN Summary 191

Summary 191

Chapter 13 Routing Choices 193

Which Routing Protocol? 194

    How Fast Does the Routing Protocol Converge? 194

    Is the Routing Protocol Proprietary? 196

    How Easy Is the Routing Protocol to Configure and Troubleshoot? 197

    Which Protocol Degrades in a Way That Works with the Business? 198

    Which Protocol Works Best on the Topology the Business Usually Builds? 199

    Which Protocol is Right? 200

IPv6 Considerations 202

    What Is the Shape of the Deployment? 202

    How Does Your Deployment Grow? 202

        Topological Deployment 203

        Virtual Topology Deployment 203

    Where Are the Policy Edges? 203

    Routing Protocol Interaction with IPv6 204

        IS-IS Interaction with IPv6 204

        OSPF Interaction with IPv6 205

        EIGRP Interaction with IPv6 206

Deploying BGP 206

    Why Deploy BGP? 207

        Complexity of Purpose 207

        Complexity of Place 208

        Complexity of Policy 208

    BGP Deployment Models 209

        iBGP Edge-to-Edge (Overlay Model) 209

        iBGP Core 210

        eBGP Edge-to-Edge (Core and Aggregation Model) 211

Summary 212

Chapter 14 Considering Complexity 213

Control Plane State 213

    Concepts of Control Plane State 214

    Network Stretch 215

    Configuration State 217

Control Plane Policy Dispersion 218

Data Plane State 220

Reaction Time 223

Managing Complexity Trade-offs 225

Part V Current and Future Trends

Chapter 15 Network in Motion 227

The Business Case for Mobility 228

    A Campus Bus Service 228

    A Mobile Retail Analysis Team 229

    Shifting Load 230

Pinning the Hard Problems into Place 230

    Mobility Requires State 231

    Mobility Requires Speed 231

    State Must Be Topologically Located 232

    State and the Network Layers 233

IP-Centric Mobility Solutions 234

    Identifier-Locator Network Protocol (ILNP) 235

    Locator Identifier Separation Protocol (LISP) 237

    Mobile IP 238

    Host Routing 239

    Mobile Ad-Hoc Networks (MANET) 240

    Dynamic DNS 242

    Final Thoughts on Mobility Solutions 243

Remote Access Solutions 244

    Separate Network Access from Application Access 244

    Consider Cloud-Based Solutions 245

    Keep Flexibility as a Goal 246

    Consider Total Cost 248

    Consider Making Remote Access the Norm 248

What Solution Should You Deliver? 249

Chapter 16 On Psychologists, Unicorns, and Clouds 251

A Cloudy History 252

This Time It’s Different 254

What Does It Cost? 255

What Are the Risks? 256

What Problems Can Cloud Solve Well? 257

What Services Is Cloud Good at Providing? 258

    Storage 258

    Content Distribution 259

    Database Services 260

    Application Services 260

    Network Services 260

Deploying Cloud 261

    How Hard Is Undoing the Deployment? 261

    How Will the Service Connect to My Network? 261

    How Does Security Work? 262

    Systemic Interactions 262

Flying Through the Cloud 262

    Components 263

Looking Back Over the Clouds 264

Chapter 17 Software-Defined Networks 265

Understanding SDNs 265

    A Proposed Definition 265

    A Proposed Framework 266

        The Distributed Model 267

        The Augmented Model 268

        The Hybrid Model 269

        The Replace Model 271

        Offline Routing/Online Reaction 272

        OpenFlow 274

        Objections and Considerations 276

    Conclusion 281

Software-Defined Network Use Cases 281

    SDNs in a Data Center 281

        What OpenFlow Brings to the Table 281

        Challenges to the OpenFlow Solution 283

    SDNs in a Wide-Area Core 283

Final Thoughts on SDNs 285

Chapter 18 Data Center Design 287

Data Center Spine and Leaf Fabrics 287

    Understanding Spine and Leaf 288

    The Border Leaf 291

    Sizing a Spine and Leaf Fabric 291

        Speed of the Fabric 291

        Number of Edge Ports 292

        Total Fabric Bandwidth 293

    Why No Oversubscription? 294

The Control Plane Conundrum 295

    Why Not Layer 2 Alone? 295

    Where Should Layer 3 Go? 296

        Software-Defined Networks as a Potential Solution 298

Network Virtualization in the Data Center 299

Thoughts on Storage 299

Modularity and the Data Center 300

Summary 301


9781587143755   TOC   3/12/2014


Published by Cisco Press (April 4th 2014) - Copyright © 2014