Kafka is the platform for stream processing. The real power of Kafka comes in the scenario of distributed computing where Kafka logs can be considered as a single source of ordered-truths. Those logs can be consumed by all the relevant consumers in their own pace and own time. (within the retention period, of course). Here is a great LinkedIn blog on log, which is the core of Kafka platform. In summary, Kafka has a number of clustered broker nodes (which actually store message inside various topics which can be partitioned (default is 3)).