Replicating across regions

In Chapter 5, Turning the Cloud into the Database, we covered the details of how an autonomous service leverages cloud native datastores to create an inbound bulkhead to protect it from disruptions in upstream services. In this section, we cover how regional data replication works at the change event level to protect autonomous services from regional disruptions. Specifically, we investigate multi-master and round-robin replication. But first, we need to address the advantage of change-event versus domain-event replication.

Change event vs domain event replication

Change events are emitted by a datastore's CDC stream as an autonomous service performs its fine-grained inserts, updates and deletes. Domain events are emitted by an autonomous service to communicate to the system that an important course-grained state change has occurred. Autonomous services may publish change event to the bus to create a complete audit trail in the event lake, but they primarily...