Guaranteed messages delivery:Challenges of transactional Outbox & Inbox patterns
summary
Java Spring Boot systems using Kubernetes and an optical‑fiber program report 500+ microservices across 50 teams, and solve atomicity by writing events to an outbox table then running an outbox process that publishes and marks events as sent. The inbox counterpart reads from brokers into the DB before business logic, relying on at‑least‑once delivery and publisher‑assigned unique IDs to deduplicate. The talk covers push vs pull brokers, REST intermediaries, scalability, housekeeping, backups and monitoring.
AI-written summary · about
chapters (14)
- 0:00Guaranteed message delivery
- 1:54Publish-subscribe pattern
- 3:23Event-driven challenges
- 4:28Transactional Outbox pattern
- 5:55Outbox failure handling
- 7:30Two Generals problem
- 8:56Transactional Inbox pattern
- 10:31Duplicate messages
- 11:53Broker limitations
- 13:06Pull vs push brokers
- 14:10Exactly-once delivery
- 15:03Scaling challenges
- 16:05Backup and monitoring
- 16:51Summary
related talks
01h 01mBeyond Pub/Sub - Advanced Messaging Patterns
01h 03mBeyond Pub/Sub - Advanced Messaging Patterns
34mAtomicity and Consistency Across Databases and Messaging Systems
48mOutbox Pattern: When That Shot at the API Isn’t Enough
01h 02mMessaging Patterns for Modern Software Solutions
47mReliably Sending Events to Apache Kafka: From CDC to the Transactional Outbox Pattern
53mEffective Microservice Communication and Conversation Patterns
59mSucceeding at Reactive Architecture