saga-orchestration

Saga orchestration coordinates distributed transactions by defining ordered forward actions across services and triggering reverse-order compensation rollbacks if a step fails. It ensures system consistency without two-phase commit by persisting saga state and managing per-step timeouts for recovery.

Design compensation logic by building reversible actions that undo completed work in reverse order. The saga pattern requires defining explicit compensation paths alongside forward actions, ensuring that completed steps are safely rolled back when a later step fails.

Recover stuck saga states and handle failed compensations by configuring per-step timeouts, retrying failed actions, and routing unrecoverable events to a dead-letter queue. Saga state persistence and correlation IDs track workflow progress to identify and safely recover stuck distributed transactions.

Yes, idempotent step handlers are required for orchestration and choreography workflows. Because saga recovery relies on retrying failed forward actions and compensations, idempotent handlers ensure that duplicated retries do not cause inconsistent state or duplicate side effects across distributed microservices.

Use saga orchestration instead of event-driven choreography when a distributed workflow requires explicit centralized coordination of step ordering, per-step timeouts, and direct management of compensation paths. Orchestration provides clearer state persistence and failure recovery boundaries compared to decentralized event-driven choreography.