Call me maybe - communication between microservices
summary
Compares REST, gRPC and event buses with a hands on Go demo that creates and fetches a simple "human" object up to 500k entries, measures local benchmarks and container behaviour, and reports gRPC 4x faster locally ≈10x in containers; shows protobuf codegen avoids manual JSON marshalling; surveys Kafka RabbitMQ SQS Cloud Pub/Sub and NATS and states production use of NATS as event bus; gives concrete REST design rules, versioning guidance, and tradeoffs for delivery guarantees and ordering.
AI-written summary · about
chapters (18)
- 0:00Introduction and speakers
- 1:29Monolithic applications explained
- 2:33Evolution of tech stacks
- 6:53Modern system requirements
- 7:58REST APIs overview
- 9:24GraphQL and gRPC explained
- 11:11Event buses and messaging
- 12:13Live demo setup
- 13:37REST API implementation
- 17:15gRPC implementation
- 19:44Benchmark results comparison
- 20:49REST vs gRPC pros and cons
- 22:12Designing good REST APIs
- 23:34Decomposing microservices
- 23:55Event bus technologies
- 27:02Form3's hybrid architecture
- 27:43Key conclusions and advice
- 28:47Q&A session begins
related talks
31mBuilding a microservices architecture with Go
55mForget about HTTP
24m"Building Scalable Microservices with Go"
49mCommunication in a microservice architecture
46mgRPC vs REST: let the battle begin!
41mForget about HTTP
28mMicroservices – Boyan Dimitrov
49mBuilding fast microservices with binary protocol - introduction to gRPC