Being able to control which table the data gets inserted into by using the JSON itself is a lot easier for demonstration purposes and cases where editing the JSON is doable but interacting with Kafka isn't.

This would also provide an alternative to using topic names with dots in them as they cause a warning to be logged by Kafka whenever those are created (some conflict with underscores in metric names).

This would require an auxiliary parameter to control which field from the JSON is used to extract the target table.