🧩 Outbox Pattern (Transactional Outbox)

Outbox Pattern is a representative Integration pattern to resolve inconsistency between data consistency and event delivery in inter-service coordination.

✅ Overview

Takes a structure where (A) DB Update and (B) Event Recording are performed simultaneously within application's local transaction, and then events are delivered to outside from buffer (Outbox table).

✅ Problems Addressed

• DB update succeeded but event transmission failed.
• Event was sent but DB update was rolled back.
• Want to avoid distributed transaction (2PC).
• Cannot maintain strong consistency between microservices.

Outbox Pattern prevents discrepancy by confining everything in local DB transaction.

✅ Basic Philosophy & Structure

• Outbox table is buffer for "unsent events".
• Dispatcher reads Outbox periodically and sends to Kafka / SNS / etc.
• Delete or mark Outbox record as done after success.

✅ Suitable Applications

• Data consistency between microservices is important.
• Business systems centered on CRUD + Event coordination.
• Cases wanting to leave audit / change history.

❌ Unsuitable Cases

• Generating massive events at ultra-high frequency (Outbox becomes bottleneck).
• Workloads where latency is extremely severe.

✅ Related Styles

• Change Data Capture (CDC)
• Event-driven Architecture (EDA)
• Saga Pattern (Especially Orchestration)

✅ Summary

Outbox Pattern is the most practical consistency assurance pattern in microservices era, and is a standard method to avoid discrepancy between DB update and event delivery.