The Architecture of Event Sourcing

The last lesson showed the core loop: append events, replay them, derive state. This lesson covers the parts you need to run that loop in production. Those parts are the event store, concurrency control, snapshots, projections, and schema evolution.

The Event Store

An event store is the database that holds your events. Writing and loading one aggregate need only two operations.

• Append new events for one aggregate.
• Read all events for one aggregate, in order.

That is the whole write and load path. There is no update operation and no delete operation.