Integration patterns & trade-offs for loosely coupled systems
summary
Argues that message-based integration yields location and temporal decoupling but has measurable costs: marshalling, latency and operational complexity. Demonstrates AWS patterns with EventBridge API destination (requires an SQS buffer to avoid DoS), EventBridge Pipes (pulls from SQS to respect destination rate limits), and SQS FIFO message groups (preserve order per group while allowing parallelism). Shows retries causing infinite loops (batch limit 10 vs an 11-element payload), TTL/dedup windows (5 minutes) and dead‑letter/archive replay as practical mitigations.
AI-written summary · about
chapters (11)
- 0:00Introduction to integration
- 3:39Integration patterns overview
- 7:10Coupling in distributed systems
- 11:27Architecture evolution example
- 17:14Control flow and flow control
- 25:08Message ordering trade-offs
- 32:37Error handling and retries
- 37:14Infinite loops in event-driven systems
- 41:49Messaging vs streaming
- 42:54Cloud automation importance
- 44:39Conclusion and resources
related talks
47mEverything Loosely Coupled - Advanced Integration Patterns and Their Trade-Offs
30mApplication integration patterns for microservices
48mIntegration- and conversation-patterns for microservices
42mIntegration and Conversation Patterns in SaaS Applications
44mThinking Asynchronously: App Integration Patterns for Microservices
42mIntegration and Conversation Patterns in SaaS Applications
44mBuilding Distributed Applications with Event-driven Architecture
59mIntegration Patterns for Microservices