What you need to know about the comeback of RPC
summary
Argument: RPC resurfaces because transport and tooling solve old costs, not because distributed computing suddenly changes. Demonstrated a .NET Core gRPC workflow: write a .proto, generate code, run a simple gRPC with HTTP/2 and protobuf example that does an add RPC and a driver-position server-client streaming flow which completes and returns a final response. Cited scale drivers — 1,200 req/s now and 10M/s planned; two billion containers weekly — and claims lower latency, multiplexing and reduced connection churn.
AI-written summary · about
chapters (23)
- 0:00Introduction and speaker intro
- 2:05History of RPC from 1230s
- 5:43RPC origins in ARPANET 1976
- 7:54Gold standard RPC by Waldo and Nelson
- 9:05Early criticism of RPC
- 10:30RPC as a sacred cow in 1988
- 11:37RPC vs. object-oriented systems
- 13:53SOAP and XML era
- 14:36REST dominance and its drawbacks
- 15:23Forces behind RPC resurgence
- 16:47Containers and network efficiency
- 18:13Performance outliers and RPC benefits
- 18:58Uber’s shift from REST to RPC
- 19:42Go language and RPC adoption
- 20:28RPC for CPU and battery efficiency
- 21:11HTTP/2 and RPC advantages
- 22:57Modern RPC vs. traditional RPC
- 23:17gRPC as an RPC example
- 25:30gRPC demo walkthrough
- 34:11REST vs. RPC fallacies
- 40:16When to choose RPC over REST
- 41:40Perpetual motion machine analogy
- 42:52Conclusion and Q&A
