Does Confluent have any guidance on kubernetes deployments of typical Kafka producer/consumer applications that do not use a pipeline framework like Flink? I am routinely wondering about whether or not to group my applications together in one deployment and namespace or whether or make them totally independent, etc.
If you need network requests between containers, they can be in the same pod or namespace (e.g. Kafka Streams Inteactive Queries, or Kafka Connect clusters), but if not (e.g. regular producer/consumer), then it doesn’t really matter, as consumer-groups would only communicate with the brokers, not with other clients.