Stream-style messaging development
summary
Claims that combining async servlet/JAX‑RS handlers, backpressure (reactive streams) and stream brokers yields much higher end‑to‑end throughput than classic synchronous JMS stacks and DB polling, and shows microbenchmarks to prove it. Uses Rabbit, ActiveMQ, ZeroMQ and Apache Kafka writes append only to segment files and compares serialization: Java serialization is fastest CPU wise but largest, MessagePack cuts payload ~15–20% vs JSON. Demonstrates lost/duplicate message risks, file‑watch pitfalls, and that reactive pull flow prevents overload in practice.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:57Integration problems
- 7:11Async processing in Java
- 10:45REST and data formats
- 17:13JMS drawbacks
- 24:18System integration methods
- 31:06Reactive streams and Kafka
- 36:31Q&A
related talks
41mForget about HTTP
01h 28mHigh Performance Stream Processing
50mAddressing Recent Events
42mIt's like 10,000 streams when what you need is a queue
54mReacting to an Event-Driven World
53mSpring and Messaging: The next development generation
48mSpring for Rabbit Super Stream and SQL filters
49mReactive Spring