If distributed transactions are not feasible, which pattern helps maintain consistency across services?

• A. Global locking across services.
• B. Immediate strong consistency with synchronous commits.
• C. Ignoring cross-service consistency.
• D. Eventual consistency with conflict resolution and sagas.

Answer: (D)

When distributed transactions aren’t feasible, coordinating updates across services becomes challenging. Global locking across services or forcing immediate strong consistency with synchronous commits would require distributed ACID-style coordination, which severely hurts scalability and availability in a microservices setup. Ignoring cross-service consistency would leave the system in a brittle, inconsistent state whenever a partial failure occurs.

The pattern that fits this situation is eventual consistency with conflict resolution and sagas. In a saga, each service performs its local transaction and records the outcome, while a central orchestrator or a set of coordinated events handles the next steps. If something later fails, compensating actions are triggered to undo or offset earlier steps, guiding the system toward a consistent state over time. This approach lets services evolve independently, keeps latency reasonable, and provides a resilient way to recover from partial failures.

For example, in an order process, one service might reserve inventory while another processes payment. If the payment fails after inventory is reserved, a saga would trigger a compensating action to release the inventory. Over time, all parts of the system converge to a consistent outcome without locking resources across services.