Reactive web services
summary
Live Java examples use Spring and Project Reactor to show practical gains: moving blocking I/O to async servlets or wrapping with Mono and using subscribeOn(elastic) restores throughput (benchmarked with siege: 8 TPS with an 8-thread Tomcat limit, 16 TPS after moving work off request threads). Switching to Flux and streaming JSON avoids buffering nine 1GB documents (no 9GB heap), and backpressure via request(n) (and a JDBC iterator/hibernate stream demo) lets the producer emit only as much as the consumer can handle.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 2:02Reactive web services
- 6:56Threat limit
- 7:28Thread limit
- 8:24Latencies
- 13:32Mono
- 14:39Project Reactor
- 53:44Batch Bulk Scheduler
- 54:22Operators
