For MySQL, changes can be streamed from a secondary, i.e. the topology is this:

Primary - (MySQL replication) -> Secondary -> Debezium

Some background: https://speakerdeck.com/gunnarmorling/practical-change-data-streaming-use-cases-with-apache-kafka-and-debezium-qcon-san-francisco-2019?slide=48

Reasons for doing this (would be nice to have this in the docs somewhere):

• Not wishing to connect DBZ to Primary
• Cannot change binlog mode of existing DB to "row"
• Multiple DBZ connectors, without adding multiple connectors/load to the Primary