Netflix’s Secret to Safe Automation at Scale
summary
Netflix reports 3,000+ JVM applications, 5,000 repositories and ~1,000 Java engineers and explains how automated changes go from weeks to hours: a custom Nebula Gradle distribution plus a resolution rules plugin apply substitution, replacement and alignment rules to move most of the fleet within hours (used on log4j and guava), a nightly RocketCI job opens PRs across every repo, and an Astrid artifact index plus campaign UI, predictive test selection and remote test execution raise daily deployment frequency to ~60%.
AI-written summary · about
chapters (15)
- 0:00Intro
- 5:00Example: Delivering code is hard
- 6:51How to improve delivering changes at scale?
- 7:39Build options – Paved path
- 10:57What about transitive dependencies?
- 17:07Artifact observability
- 21:17Distributed refactoring
- 24:30Code changes are crafted with many tools
- 25:17Continuous delivery
- 28:43Testing & validation
- 35:46Shift left testing
- 42:18Better artifact observability
- 42:53Recap
- 43:42Takeaways
- 44:24Outro
related talks
37mNetflix + Gradle A Journey in Developer Productivity
25mJourney to Automated Releases
42mBuilding and Deploying Netflix
58mBuilding a Culture of Continuous Delivery
35mHow Jenkins Builds the Netflix Streaming Service
42mDependency Updates: Before and After
31mHow Netflix Thinks About DevOps
50mHow Netflix Is Upgrading to Spring Boot 3