Managing Contextual Metadata Across a gRPC Ecosystem
summary
Netflix describes running thousands of gRPC servers handling millions of requests per second and uses protobuf v3 with custom extensions and a central registry to define ownership and strict schemas. Context is serialized to binary, carried in gRPC metadata headers and returned as field-mask trailers, and transported by custom gRPC interceptors integrated into a paved-path framework. Observability relies on tracing spans and metrics to track keys, header sizes, and lost propagation, with limits and alerts to prevent header-limit failures.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:02Contextual metadata use cases
- 4:11Lessons from prior architecture
- 6:15Protobuf data management
- 8:41Context lifecycle and threading
- 11:30gRPC interceptors for transport
- 14:34Observability for success
- 15:57Architecture summary and benefits
- 17:02Q&A: Lost context handling
- 19:16Q&A: Streaming and context
- 19:57Q&A: Context in other languages
- 21:02Q&A: Metadata conflicts in fan-out
- 22:25Q&A: Pub/sub vs APIs propagation
- 23:05Q&A: Header size limits
- 24:10Q&A: Failure injection via context
- 25:34Q&A: Open sourcing solution
- 26:16Q&A: Trailers vs response data
- 27:41Q&A: Upstream data decisions
- 29:05Q&A: Protocol vs metadata
related talks
28mHow Netflix Makes gRPC Easy to Serve, Consume, and Operate - Benjamin Fedorka, Netflix
22mHow Netflix Leverages gRPC for Resilience at Scale
33mgRPC For Microservices: Service-mesh and Observability
59m.NET gRPC - deep dive
35mThe Story of why we migrate to gRPC and how we go about it
29mThe Story of Why We Migrate to gRPC and How We Go About It
01h 00m.NET gRPC - Deep Dive
16mKeynote: Overview of gRPC