We need to replicate 6 Oracle tables from same schema to Kafka for near real-time ingestion of incremental data. Should we replicate to 1 single Kafka topic or to 6 different topics (1 topic per table)? Is there a calculation we can use based on the server RAM that replicat has available? Obviously 1 table per topic will not work if we add numerous tables to replicate in the future because of the limited RAM available in the server. At the same time, having 1 topic for all tables limits performance because of lack of parallelism. Any design advice is appreciated.