Until you break Apache Kafka you don't know how fragile it is. Why would you deliberately break Kafka? Because you want to learn what can go wrong and how to prevent it. When Kafka breaks in production you'll want to fix it as fast as possible. It's cheaper to invest into learning how to chaos test Kafka than it's to hire lawyers because you broke an SLA. The purpose of chaos testing is to see what happens when a service fails. Instead of assertions, chaos tests use monitoring solutions to prove correctness.