Pros and Cons of Having Multiple Topics vs. One Topic per Consumer Group

Hey all! I have a question for you, is about having more or less consumer groups. I know that a consumer group can handle the consumption of multiple topics, but, what are the benefits of it?
I know one con: whenever I need to reset the offset for one topic within the consumer group, kafka forces me to stop the consumption from all the topics in the consumer group, and this is a big pain for us.
So, the question is, how bad would it be to follow a pattern like “only one topic for consumer group”? Would that increase the load on the cluster? Which are the risks? :pray:

Hi Odin! I went away to find an answer to your question and I got this response:

“I think his suggestion would in fact be the recommended route since partitions are the unit of parallelism. He’d still have to have a configured strategy for rebalances for when one consumer in the group goes offline. some more info here https://www.oreilly.com/library/view/kafka-the-definitive/9781491936153/ch04.html

You can also get that book at https://www.confluent.io/resources/kafka-the-definitive-guide-v2/

> how bad would it be to follow a pattern like “only one topic for consumer group”?
Not bad at all. Feel free to follow the one topic per consumer group model––that’s completely reasonable.

> Would that increase the load on the cluster?
Yes, but you’d be increasing load anyway whenever you’re adding consumers. Generally, each broker can handle thousands of concurrent connections, so when you get up to that amount, just be aware.

> Which are the risks?
None that I can really think of. It’s a pretty standard model of operation. The reason you’d want to have multiple topics per consumer group is for specific business logic where both topics are co-partitioned and you need to receive updates concurrently on the same consumer. :woman-shrugging:

Thanks a lot for the detailed answer !! very useful one :slightly_smiling_face: