Patterns of Event Driven Architecture
summary
Compares events vs messages and shows concrete patterns: event forwarding (synchronous publish/persisted messages, client-ack, last-participant-support) guarantees no loss but costs 3–4x throughput (non-persistent ~12k/s vs persisted ~3–4k/s) and introduces duplicates; the domain broker pattern partitions brokers by domain to avoid a single write dependency; two state approaches (push to an order-owned queue vs a centralized state tracker that pulls events); and the multi-broker pattern (round-robin to multiple brokers to raise capacity at the expense of FIFO ordering; mentions Kafka, JMS, SQS).
AI-written summary · about
chapters (8)
- 0:00Introduction to EDA
- 5:26Patterns and differences
- 8:25Events vs. messages
- 12:36Event forwarding pattern
- 21:36Domain broker pattern
- 30:04Broker quantum explained
- 35:43Workflow state patterns
- 46:24Multi-broker pattern
related talks
27mPatterns for Event-Driven APIs
31mDemystifying Event-Driven Architectures
48mEvent-Driven Architecture – Benefits and Pitfalls
26mEvent Driven APIs
57mEvent Driven Collaboration
41mGetting started with Event-Driven Architectures in 2023
52mSo You Want to Build An Event Driven System?
49mEvent Driven Collaboration