Event Sourcing – the what, why & how - Anita Kvamme - NDC Oslo 2024

Event Sourcing stores the complete history of domain events instead of just current state, providing an audit log and enabling time travel debugging

Four key building blocks:

• Event Store (stores immutable events)
• Command Handlers (process user actions)
• Read Models (project events into queryable views)
• Reactors (handle automation and integration)

Benefits:

• Never loses data or historical context
• Easier to extend with new features and event types
• Natural fit for task-based user interfaces
• Provides audit logging and debugging capabilities
• Good for distributed systems and modular architectures

Best suited for:

• Business domains requiring audit trails
• Complex automation scenarios
• Systems needing historical replay
• Task-based interfaces rather than CRUD
• Distributed architectures

Challenges:

• Learning curve with new way of thinking
• Eventual consistency considerations
• Cannot modify historical events directly
• Need to handle compensating actions for corrections
• More complex architecture with multiple moving parts

Implementation tips:

• Keep aggregates and event streams small
• Use compensating events for corrections
• Separate internal from external events
• Design read models for specific UI needs
• Consider eventual consistency requirements carefully

Works well with:

• Domain-Driven Design (DDD)
• Event Storming for modeling
• Microservices architectures
• Event-driven architectures
• Task-based user interfaces