Reactive Services in practice with Webflux, Kafka and DynamoDB
summary
Shows a production reactive pipeline built with Spring Webflux, Kafka and DynamoDB that handles internet banking at four million users with 318k simultaneous online, ~400 transactions per second and $130M processed in 2018. Uses Mono and Flux with backpressure, per-transaction Kafka topics, retries and dead-letter topics, and saga orchestration for async steps. Stress tests run ~80k simulated users for an hour with 30 GB of request payload and 21 GB responses, typical latency ~200 ms and max under the 5 second requirement.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:40DBServer
- 1:47The problem
- 2:36Reactive Programming
- 3:30The capabilities of Spring Webflux is
- 4:15How Spring Webflux MONO works
- 5:34Spring Webflux - FLUX
- 7:50Apache Kafka numbers
- 8:15Amazon DynamoDB
