Event sourcing with Marten

Transcript

1. Event sourcing with Marten Willem Meints
2. Chief AI Architect @ Aigency @willem_meints Willem Meints
3. • What is event sourcing? • Building your first event sourced logic • Adding Projections • Other concerns Agenda
4. What is event sourcing
6. Profile service - Domain model
7. A traditional approach API Command handler Domain class C Repository State Q C
8. An event sourced approach API Command handler Domain class C Repository C E E E Projection E Repository S S Q S
9. • In case you need an audit log of every action • Point-in-time restore of the state based on past events • Scale read operations and write operations separately Why would you use event sourcing?
10. • It doesn’t feel natural at first. • It is more work to build an event sourced system. • Performance profile can be harder to manage Event sourcing comes at a price!
11. Building your first event sourced logic
12. • Document database • Uses Postgresql database • Supports event sourcing Introducing Marten into the picture
13. Storing events with Marten
14. Adding Projections
15. • Whenever we want to read, we need to replay events. • Projection provided by aggregate often not enough. • Have you tried combining several aggregates? Why do you need projections?
16. Synchronous projections • Great for generating views • Run in the same transaction • Can generate multiple documents • Run in separate transaction • At least once processing • Useful for integrating with other data outputs Two types of projections in Marten Asynchronous projections
17. Building a projection
18. • Optimistic concurrency control • Optimizing performance with snapshots • Integrating with an event bus Other concerns
19. • Event sourcing can be quite useful in more complex domains • Marten provides an excellent approach to it • Please be aware of the costs Summary
20. Follow the tutorial: https://github.com/wmeints/eventsourcing-and-cqrs-with-marten Get the demo: https://github.com/wmeints/event-sourcing-with-marten Try it yourself!