Kill It with Fire: Manage Aging Computer Systems (and Future Proof Modern Ones)

Description

Introduction Chapter 1: Time Is a Flat Circle Leveraging What Came Before The User as the Market Co-Creator The Mainframe and the Cloud The Flat-Rate Internet Migrating for Value, Not for Trends Chapter 2: Cannibal Code Alignable Differences and User Interfaces Unix Eats the World Inheritance Paths Leveraging Interfaces When Approaching Legacy Systems Beware Artificial Consistency Chapter 3: Evaluating Your Architecture Problem 1: Technical Debt Example: The General Ledger Problem 2: Performance Issues Example: Case Flow Management Problem 3: Stability Issues Example: Custom Configuration Stages of a Modernization Plan No Silver Bullets Full Rewrite Iteration in Place Split in Place Blue-Green The Hard Cutoff Putting It Together Chapter 4: Why Is It Hard? The Curse of Hindsight Easy and Also Impossible Overgrowth: The Application and Its Dependencies Shifting Vertically: Moving from One Abstraction Layer to Another Shifting Horizontally: Moving from One Application to Another Shifting from Client to Server Shifting Down the Dependency Tree Cutting Back the Overgrowth Automation and Conversion Transpiling Code Static Analysis A Guide to Not Making Things Harder Chapter 5: Building and Protecting Momentum Momentum Builder: The Bliss of Measurable Problems Anatomy of the Measurable Problem Momentum Killer: The Team Cannot Agree Step 1: Define a Scope Step 2: Check for Conflicting Optimization Strategies Step 3: Perform Time-Boxed Experiments Momentum Killer: A History of Failure Momentum Builder: Inspiring Urgency Protecting Momentum: A Quota on Big Decisions Protecting Momentum: Calculating Opportunity Costs The Cost of Not Adding New Features The Cost of Not Fixing Something Else The Cost of Not Deprecating in Favor of a Different Solution Chapter 6: Coming in Midstream Finding the Bleed Mess: Fixing Things That Are Not Broken Figuring Out Whether Something Needs to Be Fixed But . . . What About Conventions? When Does Breaking Up Add Value? The Compounding Problem: Diminishing Trust Solution: Formal Methods Mess: Forgotten and Lost Systems The Compounding Problem: Crippling Risk Avoidance Solution: Chaos Testing Mess: Institutional Failures The Compounding Problem: No Owners Solution: Code Yellow Calling a Code Yellow Running a Code Yellow Mess: Leadership Has Lost the Room The Compounding Problem: Self-Sabotaging Teams Solution: Murder Boards Stopping the Bleed Chapter 7: Design as Destiny Designing Technical Conversations How to Run a Design Exercise More About Follow-ups: Why vs. How Some Useful Design Exercises for Engineering Teams Exercise: Critical Factors3 Exercise: The Saboteur4 Exercise: Shared Uncertainties5 Exercise: The 15 Percent6 Exercises Specifically for Decisions Exercise: Probabilistic Outcome-Based Decision-Making Exercise: Affinity Mapping Team Structure, Organization Structure, and Incentives Individual Incentives Minor Adjustments as Uncertainty Organization Size and Communication Manager Incentives Designing a Team: Applications of Conway’s Law Reorgs Are Traumatic Finding the Right Leadership Exercise: The Smallest Testable Unit Structuring the Team to Account for Past Failure Exercise: In-Group/Out-Group Takeaways Chapter 8: Breaking Changes Being Seen Who Draws the Line? Building Trust Through Failure Breaking Change vs. Breaking Why Break Things on Purpose? Projecting Impact The Kill Switch Communicating Failure Failure Is a Best Practice Chapter 9: How to Finish Revealing Assumptions Approach 1: Success Criteria Example: Adding Continuous Integration/Continuous Deploy Approach 2: Diagnosis-Policy-Actions Example: Upgrading a Database Comparison Marking Time Postmortems on Success Postmortem vs. Retrospective Running Postmortems The Tale of Two War Rooms Working Groups vs. Committees Success Is Not Obvious If Undefined Chapter 10: Future-Proofing Time Unescapable Migrations Failing Gracefully Less Time Between Upgrades, Not More A Word of Caution About Automation Building Something Wrong the Correct Way Feedback Loops Don’t Stop the Bus Conclusion Index