100% Fast and Resilient Microservice with RSocket
summary
A Spring Boot WebFlux microservice demo (Petchkin → BigBrother → AgentSmith) shows HTTP strategies that fail under load: drop, retry, local rate limiting and a distributed attempt using hazelcast + bucket4j both break when random latency and multiple Petchkin instances appear. Replacing HTTP/WebClient with RSocket Java and enabling RSocket leasing (server-driven concurrency, Netflix concurrency-limit for capacity calc) prevents queue blowups, preserves request/response patterns, and stabilizes the chain; live demo runs, and built-in load balancing plus resumability are shown as next steps.
AI-written summary · about
chapters (18)
- 0:00Start
- 9:41Prerequisites
- 18:14Microservices System
- 34:04Why Not To Use Thread Pulls
- 38:40Processing Properties
- 46:55Reactor Pool
- 48:05Dashboard
- 1:02:47Rate Limit
- 1:02:58Rate Limiting
- 1:17:38Synchronization
- 1:19:22Base Configuration
- 1:30:25Summary
- 1:31:01Why Concurrency Limiting Is Better than Rate Limiting
- 1:38:02Flexibility
- 1:38:39Reactive Streams
- 1:45:16Netflix Concurrence Limit
- 1:49:17Load Balancing
- 1:50:23Predictive Load Balancing
related talks
42mRSocket - Future Reactive Application Protocol
40mReactive message driven service 2 service communication with RSocket
28mGetting Started With RSocket
01h 01mThe RSocket Protocol — The Future of Reactive Applications
44mDeep Dive talk | Resilience in Microservices with Congestion Control
27mEnabling real time collaboration with RSocket
38mResilience patterns in practice
53mMultiplayer Pac-Man with RSocket