grpcgi: Enabling Universal Web Serving, Native mTLS, and xDS on the gRPC Mesh
summary
grpcgi inverts the transport so Envoy speaks gRPC to an in‑process server that translates into ASGI/Rack, letting unmodified web apps inherit SPIFFE identity, native mTLS, xDS routing and gRPC load balancing. Prototype is an Envoy extension (Envoy extension via proxy‑wasm with a native C++ filter) plus a language‑neutral, streaming proto. The streaming gRPCGI proto with websocket/SSE support carries headers, body chunks and trailers; an alpha ASGI server runs Django unchanged and a live WebSocket echo demo works over xDS + mTLS.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:17HTTP in gRPC mesh
- 5:14Sidecar vs built-in gRPC
- 7:42Request flow model
- 10:55WebSocket handling
- 13:02Session resumption
- 15:34Framework integration
- 17:20Q&A
related talks
23mA Simplified Service Mesh With gRPC
33mgRPC For Microservices: Service-mesh and Observability
20mProxyless Service Mesh with gRPC
23mHow Proxyless gRPC Works in a Service Mesh
19mKeynote: What's New in gRPC?
24mWhat's New in gRPC
34mgRPC Proxyless Service Mesh with Security
16mxDS in gRPC for Service Mesh
