Reactive Architecture
summary
Claims that nonblocking, event-driven design and RSocket plus reactive DB drivers let systems sustain spikes (tests with ~10,000 concurrent connections) by avoiding thread-blocking, reducing CPU context switches and preventing OOM-driven cascading failures. Describes switching from HTTP POST buffering to streaming with application/x-ndjson frames, using backpressure so consumers request batches, and reports higher throughput and lower latency in simple benchmarks compared with blocking stacks. Notes tradeoffs: library support and extra framing work.
AI-written summary · about
chapters (11)
- 0:00Reactive architecture intro
- 3:30Traditional architecture problems
- 13:28Reactive solutions overview
- 19:12What is reactive programming
- 22:27Project Reactor implementation
- 24:37Reactive database connectivity
- 26:44Transitioning to reactive
- 27:05WebSocket and AirSocket
- 32:55Reactive architecture advantages
- 33:58Fault tolerance and scalability benefits
- 34:20Q&A
