...
You must have data that is (partitionalblepartitionable?) i.e. that your data can be split and processed independently using one or more fields as partitioning keys. These fields will then be used by Kafka to determine which partitions the UDRs are assigned to.
...