Could you explain what the term 'offset' means in the context of Apache Kafka?

In Apache Kafka, the term "offset" refers to a unique sequential number of a message within a topic partition. Each time a producer sends a message to a partition, it receives the next offset in the sequence. Consumers use the offset to track which messages they have already read, so they can resume reading from the correct position after a restart or failure. Thus, the offset is a kind of pointer to the position in the message stream.