transaction-consistency-designer

To define transaction boundaries in Spring, you wrap atomic workflow steps with @Transactional while avoiding long-held transactions. This isolates multi-repository writes to enforce invariants and prevents partial commits by rolling back the entire operation on failure.

The outbox pattern ensures reliable message delivery by writing messages to a database table within the same transaction as business data. You need it when publishing messages to external systems to prevent data inconsistency if the message broker becomes temporarily unavailable.

Implement idempotency in Kotlin by tracking processed identifiers or applying deterministic operations for external calls. This prevents duplicate processing when retries occur due to network timeouts, ensuring consistent system state without repeating side effects.

Spring @Transactional propagation and isolation behave identically in Kotlin workflows. You must apply Spring-specific checks to configure propagation behavior and exception handling correctly, ensuring transactional boundaries are enforced without triggering common anti-patterns.

Common transaction anti-patterns include holding long transactions during external calls and improper exception handling. Enforcing proper transaction boundaries prevents these issues by separating external system calls from the core database transaction scope.