How is data replication performed in Apache Kafka?

Data replication in Apache Kafka is performed at the partition level of topics. Each partition has one leader and several replicas (followers). The leader is responsible for handling all read and write operations for that partition. Replicas asynchronously copy data from the leader, maintaining up-to-dateness.

If the leader fails, one of the synchronized followers automatically becomes the new leader, ensuring high data availability. Replication is configured with the replication.factor parameter when creating a topic.

Thus, replication provides fault tolerance and scalability, allowing the system to continue operating in case of individual broker failures.