In every Kafka broker, some partitions are available, either a leader or a replica of a topic.
* Every Kafka topic separated into partitions contains records in a fixed order in each of them.
* Each record in a partition is assigned and attributed with a unique offset. Multiple partition logs are possible in a single topic. Because of this facility, several users can read from the same topic at the same time.
* Topics can be parallelized via partitions, which split data into a single topic among numerous brokers.
* In Kafka, replication is done at the partition level, and a replica is the redundant element of a topic partition.
* Each partition can contain one or more replicas, and it means the partitions can contain messages that are duplicated across many Kafka brokers in the cluster.
* One server acts as the leader of each partition or replica, while the others act as followers.
* If the leader goes down in any circumstances, one of the followers takes over as the leader.
