Data Lake for Enterprises

Partitions - Kafka topic division

To achieve fault tolerance and replication, Kafka’s topic is divided into partitions and also distributed over multiple brokers. Each partition elects one server (broker or node) as the leader (known as leader) and zero or more servers follows the leader (known as followers).

The leader takes care of the read and write activities. The followers keep replicating the leader and always sync with it. As with any distributed system, if the leader fails due to any reason, one of the followers takes charge of the situation automatically and becomes the leader. This allows seamless fault tolerance and message durability in Kafka.

To divide the leader and followers throughout the Kafka cluster, Kafka makes optimal ...