What is the difference between Kafka and RabbitMQ for microservices messaging?

Apache Kafka is a distributed log designed for high-throughput, durable, ordered event streaming. Messages are persisted to disk for a configurable retention period (days or weeks), consumers track their own offset, and messages can be replayed. Kafka excels at event sourcing, audit logs, stream processing (with Kafka Streams or Flink), and scenarios requiring very high throughput (millions of messages/sec). RabbitMQ is a traditional message broker designed for flexible message routing. Messages are pushed to consumers and acknowledged — once acknowledged, they are deleted. RabbitMQ supports complex routing topologies (exchanges, bindings), dead-letter queues, and is simpler to operate at smaller scale. Choose Kafka for event streaming and replay; choose RabbitMQ for task queues and flexible routing patterns.