How Netflix Uses Java
summary
Netflix runs about 3,000 services on Spring Boot and modern JDKs (most high‑RPS services on JDK 20–23), migrates legacy libraries via patches so teams can upgrade, and uses DGS/GraphQL for UI→gateway and gRPC for server→server. Upgrading from Java 8 to 17 cuts GC CPU by ~20%; switching from G1 to generational ZGC eliminates >1s stop‑the‑world pauses and lowers IPC timeouts and error rates. Frameworks enable virtual threads (JDK24 fixes prior pinning deadlocks), Gradle bytecode rewrites handle JavaX→Jakarta, and native images remain impractical.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 14:04Java usage at Netflix
- 23:23JDK and GC upgrades
- 27:18Virtual threads and reactive
- 32:18Spring Boot Netflix
- 40:49Spring Boot 3 upgrade
- 43:20GraphQL with DGS
- 45:29GraphQL vs gRPC
- 47:37Q&A
related talks
53mHow Netflix Uses Java - 2026 Edition
50mHow Netflix Is Upgrading to Spring Boot 3
46mGoing Java::Current in Production
14mWho runs the world ? JAVA!!!!
01h 05mJava 11 to 17 - Exciting New Features
59mJava microservices from Netflix OSS to Kubernetes
41mModern Java for the Masses! Is Java Still Relevant
51mLife Beyond Java 8