Kafka is just the broker, the stage in which all the action takes place. The producers send messages to the world while the consumers read specific chunks of data. How do you differentiate one specific portion of data from the others? How do consumers know what data to consume? To understand this, you need a new actor in the play: the topics.
Kafka topics are the channels, the carriage that transport messages around. Kafka records produced by producers are organized and stored into topics.
This is a nice overview of Kafka followed by the basics of building a consumer and a producer in C#. I just wish that there was more community usage of Kafka so that the Confluent .NET driver would include some of the really cool stuff they’ve added to Kafka over the past couple of years.