nw-ddd-eventsourcing

Event sourcing and CQRS are recommended in domain-driven design for domains needing audit trails, temporal queries, and multiple read views across services. They separate write and read models to optimize command and query operations while storing state changes as immutable events.

Design aggregates as transactional boundaries that emit domain events, and build projections to translate those events into queryable read models. This pattern ensures current state is derived by replaying events while projections serve specific query needs.

Event upcasting handles schema evolution in event stores by transforming older event versions to newer schemas during replay. This pattern ensures backward compatibility when domain models evolve without breaking existing event streams.

Conflict resolution in an event-sourced aggregate involves detecting concurrent modifications when appending events to the stream. This requires optimistic concurrency control and applying business logic to merge or reject conflicting state changes.

Testing strategies for CQRS and event sourcing focus on aggregate behavior validation through event replay and verifying projection consistency. This includes testing command handlers, event emission, and ensuring read models accurately reflect the event stream.

Avoid event sourcing when your domain lacks audit requirements, has simple CRUD operations, or when temporal queries and multiple read views are unnecessary. The operational complexity of event stores and projections outweighs benefits for straightforward state management.