What Kafka Actually Is

Kafka · lesson 1 of 34 · 3 min read

Know what Kafka is, how it differs from a queue, and when reaching for it is the right call.

Open this lesson in the learning hub

Key points

  • Kafka is a durable, replayable log. Producers append records, consumers read at their own pace. Nothing is deleted on read.
  • A classic queue removes a message once it is consumed. Kafka keeps it for a retention window, so ten teams can read the same stream.
  • It is built for throughput. Writes are sequential appends to disk, which is why one cluster handles millions of records a second.
  • Reach for Kafka when many consumers need the same events, when you need replay, or when you must absorb bursts.
  • Skip it for request/response and simple task queues. A REST call or RabbitMQ is far less to operate.
  • Kafka 4.x runs on KRaft, its built-in Raft controller. ZooKeeper is gone.

Kafka is a log you replay, not a queue you drain.

Topics, Partitions, Offsets

This is a reading copy. The full lesson — with the visual explainer, the interactive lab and a Run button for the code — lives in the Kafka course, and every lesson in it is listed on the Kafka contents page.