The Story of why we migrate to gRPC and how we go about it
summary
Spotify abandons Hermes for gRPC, claims a multi‑year migration of migrating 1000+ services to cut boilerplate and tap open‑source tooling. Migration uses protobuf proto files as source of truth with central registry tooling (buff, proto‑tool), client interceptors for retries and deadlines, and XDS‑style external load balancer control plane to avoid sidecars. Progress hits measurable milestones: early 10% then ~40% of services expose gRPC, with hybrid endpoints and compatibility checks in place.
AI-written summary · about
chapters (9)
- 0:00Intro
- 1:40Background
- 3:52Hermes
- 6:15Why gRPC
- 9:03Proto Files
- 12:32Resiliency
- 15:37Deadlines
- 22:04Mascot
- 26:57Autonomous teams
related talks
29mThe Story of Why We Migrate to gRPC and How We Go About It
36mAdopting gRPC at Spotify
30mOops!...I Did It Again — I Reinvented gRPC
21mReal World Migration from HTTP to gRPC in Ruby
24mgRPC Proxyless Service Mesh
23mRunning gRPC services for serving legacy RESTful APIs on Kubernetes
33m"RPC frameworks overview"
46mIntroduction to gRPC