Kafka: Topic vs Partition data

Question

Going through Kafka documentation and various other resources, I understand that messages in Kafka are organized into topics. Also, a topic can be broken down into partitions and each partition can be hosted on a different server. This gives redundancy and scalability.

I am not sure what the word 'broken' means here. Does it mean that if messages added to a topic are, say '1 2 3 4 5 6 7', then after breaking it into partitions, we would have one partition having only a subpart of whole topic. Like one partition having '1 2 3' while another partition having '4 5 6' and yet another having just '7'. OR does it mean that every single partition has '1 2 3 4 5 6 7', meaning we have exact replicas.

Answer 1

a topic can be broken down into partitions and each partition can be hosted on a different server. This gives redundancy and scalability

Above statement refers to - Kafka topics are usually divided into number of partitions. Partitions allows to parallize the topic by splitting the data across different brokers. If a topic contains only one partition, the data resides on single brokers and will be read sequentially. If lets's say number of partition is 3, same data will be splitted into 3 partitions, each carrying different sets of events. You can read the topic in 3 parallel process, each reading from one partition. The more number of partitions you have, the more scalablility you can achieve. Yes, each partition will have only subset of data.