Reactive programming: lessons learned
summary
Reactive programming delivers massive concurrency — C10k/C100k and even reports of 20 million connections on a JVM — and examples using Reactor, RxJava, CompletableFuture, Akka, Spring Boot and ActiveMQ. It trades higher development and maintenance costs for lower hardware cost; Tomcat with 100 threads handles ~1,000 req/s by Little’s law. Debugging suffers: 200-line stack traces become meaningless, nulls and timeouts hide their source, and implicit schedulers cause overloads. Adopt only when past the Netflix point, otherwise prefer simple blocking code with explicit queues (Seda/mailboxes).
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:20Speaker Experience
- 4:37The Cost of Complexity
- 7:07Code Evolution Example
- 10:19Ubiquitous Language
- 13:28Measuring Code Quality
- 17:31Cost Analysis
- 22:21The Netflix Point
- 26:09Troubleshooting Challenges
- 32:03Best Practices
- 37:37Conclusion
related talks
56mReactive programming lessons learned
41mThe definite guide to reactive programming in Java
35m5 Reasons to Use Reactive Programming if You Are Not eBay
46mGet Reactive! Programming, Systems, and Microservices
24mBringing Reactive Programming to Java and Microservices
40mREACTIVE A New Hope! Learn From a Performance History
47mReactive Spring
01h 01mReactive web services