A reference table stores relatively stable lookup or classification values, such as marital status, occupations, education levels, and country codes. Replicating these tables usually needs less application-specific logic than replicating profile changes. section 5.1.2
Choose the replication approach according to how the source data changes:
- Use snapshot-based replication to transfer the source’s current state. section 5.1.2
- Use change data capture (CDC) to stream inserts, updates, and deletions from source-system change logs. section 5.1.2
- Use Kafka Connect configurations and connectors instead of custom update-notification services; JDBC Source Connector supports snapshots, while Debezium can provide CDC. section 5.1.2
Replicating lookup data
Source connectors can pull reference data from PostgreSQL and MongoDB into Kafka topics. Sink connectors can then read those topics and populate an MS SQL database and Amazon S3.
