I am publishing messages to Kafka which are consumed by a number of consumers. These are idempotent so it doesn't matter if one message is consumed multiple times.
However for performance reasons I don't want my (single) producer to publish a message that already exists in the queue. Let's say messages are just ID-strings so it's very easy to check if two messages are equal.
I suppose that Kafka alone isn't made to deal with this kind of performance improvement. Are there any tools or concepts to help dealing with this issue?