Surviving Continuous Deployment in Distributed Systems
summary
Live animal-shelter example shows concrete workflows: use feature toggles (state stored outside app — S3/SSM) to implement new UI features outside-in without breaking production; use expand-and-contract / parallel change to refactor contracts inside-out (add dual-format endpoints, migrate clients, then remove old); handle DB migrations with a preemptive double-writer to avoid data loss when changing types (floats → cents). Example requires seven safe releases instead of three, trading release count for zero-downtime safety and predictable rollouts.
AI-written summary · about
chapters (14)
- 0:00Start
- 4:49Architecture
- 5:25Repositories
- 6:06Contracts between distributed components
- 7:51The old way gate to production
- 14:16Feature Toggles for Feature Flags
- 18:23Example refactoring
- 22:42Inside-Out
- 28:49Isolate changes belonging to different components
- 30:24Database Trigger
- 31:16Pre-emptive double write
- 34:10New features vs Refactoring
- 34:57Example feature addition w/refactoring
- 36:38The Story Plan
related talks
26mThe Many Ways to Deploy Continuously
23mContinuous Deployment - The Last Frontier for Continuous Delivery
48mWinning at Continuous Deployment: The Art of Safe Contract Changes
34mEnabling a Continuous Deployment Mindset
46mContinuous Deployment Workflows
23mDevOps and Legacy Code
30mExpand Contract Pattern for Continuous Delivery of Databases
31mFireside Chat: Continuous Deployment