Applying Reactive Programming with Rx
summary
Argues that replacing blocking service calls with Rx Observable APIs and wrapping legacy blocking IO (JDBC, memcached) via defer + subscribeOn(IO) yields streaming items, backpressure and lower user latency; shows Apple TV rewritten three times to get it right, roughly two months ramp per engineer, and production control-plane traffic moved to observables. Demonstrates testing with virtual time test scheduler, explains bridging async→sync with toBlocking()/forEachBlocking, and a Netty vs Tomcat benchmark where Netty scales better (systems hit 1,400 threads on JVM otherwise).
AI-written summary · about
chapters (9)
- 0:00Start
- 13:04Observable Stream Model
- 21:13Apple Tv
- 23:19Error Handling
- 25:16Unit Testing
- 31:02Observable Api
- 38:20Average Latency
- 38:34Max Latency
- 39:29Thread Migrations
