skill-arch-transactional

ACID transactional consistency across microservices is maintained using patterns like Saga choreography, Try-Confirm-Cancel (TCC), and two-phase commit (2PC). These distributed transaction patterns ensure atomic updates and proper rollback when partial failures occur.

Deadlocks and concurrency conflicts in distributed transactions are prevented using locking strategies like optimistic locking, pessimistic locking, and advisory locks. Selecting appropriate isolation levels and deadlock prevention mechanisms ensures safe concurrent operations.

The outbox pattern works by writing domain state changes and event publication records within the same Unit of Work boundary. This guarantees that outbox events are published reliably exactly once, even if downstream services experience transient failures.

Idempotency keys are needed for cross-service retries when operations must produce the same outcome regardless of duplicate requests. They enable safe, repeatable operations by ensuring distributed transaction retries do not cause duplicate side effects.

Yes, this transactional Skill provides practical code examples in both TypeScript and Python. It demonstrates implementation patterns for isolation levels, Unit of Work boundaries, savepoints, and distributed transaction coordination across microservices.

Saga choreography experiences data inconsistency during retries when idempotency is not enforced at service boundaries. Implementing idempotency keys, controlled retry strategies, and safe error handling ensures cross-service Saga operations remain consistent.