Beyond the Toggle: Feature Flags as Your Migration Safety Net
summary
Explains using feature flags as migration controls with the strangler fig pattern to bifurcate traffic and measure cohorts, moving from 1% canaries to full rollout while tracking latency error rates throughput and error budget. Describes using a feature flag to route infra traffic in a Split/Harness stack, cutting cost ~80% cost reduction and shrinking a batch job from 10h→15m processing for a 50 gigabit payload, plus cautions on stale flags ownership and a tested rollback SLA.
AI-written summary · about
chapters (10)
- 0:00Welcome and Setup
- 0:43Migration Risks and Failure Modes
- 2:52Strangler Fig Pattern Basics
- 4:19Progressive Rollouts and Measuring
- 7:44What Metrics to Track
- 10:14Measuring Versus Monitoring
- 12:09Case Study Split Data Pipeline
- 14:41Feature Flag Cautionary Tales
- 17:52Action Checklist and Rollback SLA
- 18:54Wrap Up and Next Steps
related talks
11mFeature flag use cases you haven't heard about yet
30mA Tale of Two Feature Flags
56mWhen Feature flags go bad
01h 09mFeature flags: the toggle, the A/B test and the canary
41mImproving feature flag implementation to achieve flow
59mFaster & less risky releases with feature flags
41mFeature Flags: Separate Deployment from Feature Release
20mFeature Flag-Driven Development… in Reverse