Scalability Rules

Martin L. Abbott is a founding partner of AKF Partners, a growth consulting firm focusing on meeting the needs of today's fast-paced and hyper-growth companies. Marty was formerly the COO of Quigo, an advertising technology startup acquired by AOL in 2007. Prior to Quigo, Marty spent nearly six years at eBay, most recently as SVP of Technology and CTO and member of the CEO's executive staff. Prior to eBay, Marty held domestic and international engineering, management, and executive positions at Gateway and Motorola. Marty has served on a number of boards of directors for public and private companies. He spent a number of years as both an active duty and reserve officer in the US Army. Marty has a BS in computer science from the United States Military Academy, an MS in computer engineering from the University of Florida, is a graduate of the Harvard Business School Executive Education Program, and has a Doctorate of Management from Case Western Reserve University. Michael T. Fisher is a founding partner of AKF Partners, a growth consulting firm focusing on meeting the needs of today's fast-paced and hyper-growth companies. Prior to co-founding AKF Partners, Michael held many industry roles including the chief technology officer of Quigo, acquired by AOL in 2007, and the vice president of engineering & architecture for PayPal. He served as a pilot in the US Army. Michael received a PhD and MBA from Case Western Reserve University's Weatherhead School of Management, an MS in information systems from Hawaii-Pacific University, and a BS in computer science from the United States Military Academy (West Point). Michael is an adjunct professor in the Design & Innovation Department at Case Western Reserve University's Weatherhead School of Management.

Preface xv

Acknowledgments xxi

About the Authors xxiii





Chapter 1: Reduce the Equation 1

Rule 1-Don't Overengineer the Solution 3

Rule 2-Design Scale into the Solution (D-I-D Process) 6

Rule 3-Simplify the Solution Three Times Over 8

Rule 4-Reduce DNS Lookups 10

Rule 5-Reduce Objects Where Possible 12

Rule 6-Use Homogeneous Networks 15

Summary 15

Notes 16



Chapter 2: Distribute Your Work 19

Rule 7-Design to Clone or Replicate Things (X Axis) 22

Rule 8-Design to Split Different Things (Y Axis) 24

Rule 9-Design to Split Similar Things (Z Axis) 26

Summary 28

Notes 28



Chapter 3: Design to Scale Out Horizontally 29

Rule 10-Design Your Solution to Scale Out, Not Just Up 31

Rule 11-Use Commodity Systems (Goldfish Not Thoroughbreds) 33

Rule 12-Scale Out Your Hosting Solution 35

Rule 13-Design to Leverage the Cloud 40

Summary 42

Notes 42



Chapter 4: Use the Right Tools 43

Rule 14-Use Databases Appropriately 47

Rule 15-Firewalls, Firewalls Everywhere! 52

Rule 16-Actively Use Log Files 55

Summary 58

Notes 58



Chapter 5: Get Out of Your Own Way 59

Rule 17-Don't Check Your Work 61

Rule 18-Stop Redirecting Traffic 64

Rule 19-Relax Temporal Constraints 68

Summary 70

Notes 70



Chapter 6: Use Caching Aggressively 73

Rule 20-Leverage Content Delivery Networks 75

Rule 21-Use Expires Headers 77

Rule 22-Cache Ajax Calls 80

Rule 23-Leverage Page Caches 84

Rule 24-Utilize Application Caches 86

Rule 25-Make Use of Object Caches 88

Rule 26-Put Object Caches on Their Own "Tier" 90

Summary 91

Notes 92



Chapter 7: Learn from Your Mistakes 93

Rule 27-Learn Aggressively 95

Rule 28-Don't Rely on QA to Find Mistakes 100

Rule 29-Failing to Design for Rollback Is Designing for Failure 102

Summary 105

Notes 106



Chapter 8: Database Rules 107

Rule 30-Remove Business Intelligence from Transaction Processing 109

Rule 31-Be Aware of Costly Relationships 111

Rule 32-Use the Right Type of Database Lock 114

Rule 33-Pass on Using Multiphase Commits 116

Rule 34-Try Not to Use Select for Update 118

Rule 35-Don't Select Everything 120

Summary 121

Notes 122



Chapter 9: Design for Fault Tolerance and Graceful Failure 123

Rule 36-Design Using Fault-Isolative "Swim Lanes" 124

Rule 37-Never Trust Single Points of Failure 130

Rule 38-Avoid Putting Systems in Series 132

Rule 39-Ensure That You Can Wire On and Off Features 135

Summary 138



Chapter 10: Avoid or Distribute State 139

Rule 40-Strive for Statelessness 140

Rule 41-Maintain Sessions in the Browser When Possible 142

Rule 42-Make Use of a Distributed Cache for States 144

Summary 146

Notes 146



Chapter 11: Asynchronous Communication and Message Buses 147

Rule 43-Communicate Asynchronously as Much as Possible 149

Rule 44-Ensure That Your Message Bus Can Scale 151

Rule 45-Avoid Overcrowding Your Message Bus 154

Summary 157



Chapter 12: Miscellaneous Rules 159

Rule 46-Be Wary of Scaling through Third Parties 161

Rule 47-Purge, Archive, and Cost-Justify Storage 163

Rule 48-Partition Inductive, Deductive, Batch, and User Interaction (OLTP) Workloads 166

Rule 49-Design Your Application to Be Monitored 169

Rule 50-Be Competent 172

Summary 174

Notes 174



Chapter 13: Rule Review and Prioritization 177

A Risk-Benefit Model for Evaluating Scalability Projects and Initiatives 177

50 Scalability Rules in Brief 180

A Benefit/Priority Ranking of the Scalability Rules 200

Summary 202



Index 205