How Netflix Makes gRPC Easy to Serve, Consume, and Operate - Benjamin Fedorka, Netflix
summary
Netflix runs a Java gRPC framework that wires proto generation, clients, servers, caching, retries, hedging, Eureka name resolution, custom client load balancing, Spectator metrics, Zipkin tracing, and chaos/canary testing into application jars and Spring Boot actuators. It supports on-the-fly config reloads, proto lock checks, distributed EVCache and on-heap/off-heap caches, and client/server concurrency limiters across 600 applications, 1300 services, 1500 clients. All of this is maintained by a team of three engineers and is moving many features into Envoy sidecars for polyglot support.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:58Overview
- 1:56Proto specification
- 2:52Proto configuration
- 3:53Proto API documentation
- 4:44Security
- 5:28Resiliency
- 10:08Observability
- 13:11Caching
- 14:13Integrations
- 15:56Next steps
- 16:37Questions
- 18:06Batching
- 20:12Configuration
- 22:31QA
related talks
22mHow Netflix Leverages gRPC for Resilience at Scale
53mHow Netflix Uses Java - 2026 Edition
23mReducing gRPC Call Volume Through Caching and Batching
35mCaching at Netflix: The Hidden Microservice
36mHow Netflix Is Solving Authorization Across Their Cloud [I]
07mKeynote: State of gRPC
01h 29mSpring Boot and Netflix OSS
46mIntroduction to gRPC