Apache Kafka is a beautiful system. It scales well, it is stable and it provides phenomenal system architecture flexibility. After 5 years of running production Kafka clusters, I have collected a list of tips and pitfalls. Some of them were learnt the hard way. If you work in a small team rolling out Kafka to production, those might prove useful. The article assume familiarity with basic Kafka concepts, such as brokers, topics, producers and consumers. What is more, the following points should be valid for up to Kafka 2.6.0. Without further ado.