RPC Frameworks Overview
summary
Eighty slides at 30 seconds each and a migration of more than 50 services from internal REST/JSON to an RPC hub are the concrete cases here: production system uses gRPC with Protocol Buffers, Amazon SQS for queues and HashiCorp service discovery, yielding lower latency, smaller payloads (protobuf bytes vs ~25B JSON example), shared schemas for versioning, and full-duplex streams; downsides include HTTP/2 load‑balancer issues. Alternative specifics covered: Thrift, Cap’n Proto (claims zero encoding), Finagle, Dubbo and Avro, with varied language support and tooling.
AI-written summary · about
chapters (20)
- 0:00Introduction and motivation
- 1:04What is RPC and its history
- 2:12Migration from REST to RPC
- 3:59Why RPC framework
- 5:25Overview of RPC frameworks
- 5:49gRPC framework details
- 7:39Protocols and serialization
- 9:05How Protocol Buffers work
- 11:34Comparison with JSON
- 13:02gRPC code generation
- 14:05gRPC pros and cons
- 15:32Twitch TChannel framework
- 17:21Apache Thrift framework
- 20:57Thrift pros and cons
- 22:00Cap'n Proto framework
- 24:30Finagle framework
- 26:43Dubbo framework
- 28:08Apache Avro framework
- 29:16Benchmarking considerations
- 30:01Distributed Ruby
