Saga

Intent

Implement a cross-service business transaction as a sequence of local transactions coordinated with messages, including compensating actions on failure.

Use when

• A workflow spans multiple services and must handle partial failure without distributed transactions.
• You can model compensations (undo) or alternative paths explicitly.

Avoid / watch-outs

• Don’t use sagas to “fake” strong consistency; model the business reality and compensations.
• Make the saga state machine explicit and observable; implicit workflow is un-debuggable.

Skill mapping

• architecture: pick orchestration vs choreography; define the workflow states.
• resilience: idempotency, retries/backoff, and DLQ strategy for each step.
• observability: correlation IDs across steps; metrics for state transitions and time spent.
• testing: tests for happy path + compensations + duplicate delivery.