Flight of the Flux: A look at Reactor execution model
summary
Explains how Reactor Flux and Mono separate assembly time (building operator chains) from execution time so nothing happens until subscribe, illustrated by an HTTP call that only fires when subscribe runs and a take(3) that cuts emission. Shows thread control via schedulers (elastic, parallel, single) and the concrete difference between publishOn vs subscribeOn. Describes internal concurrency: work stealing via a shared queue and drain loop with an atomic guard, and performance optimizations like macro and micro fusion, plus the checkpoint stack-trace aid.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 3:24Agenda
- 4:40Assembly time
- 9:42Stack trace
- 12:01Hot and cold publisher
- 14:55Concurrent agnostic
- 16:48Schedulers
- 23:08Combining the two
- 27:34Reactive streams
- 29:32Work stealing
- 35:00Operator fusion
- 35:48Macro and microfusion
- 40:30Questions
- 42:50Async mode in the background
- 44:35Back pressure
- 47:55Mono
- 48:48Flatmap
- 49:32Micromanagement
- 50:32CPU utilization
- 50:55Parallelization
- 53:51Reactor
related talks
53mReactive Mythbusters With Reactor
01h 10mReactor in Action - Simon Baslé, Victor Grazi
29mBuilding Reactive Systems with Reactor
56mReactive programming lessons learned
51mReactor 3, the reactive foundation for Java 8 (and Spring 5)
50mThe Future is Reactive
47mReactive Spring
06mPhil Clay at SpringOne 2020