Reading the latest extended incident report from Honeycomb makes me want to never run Kafka in production.

Which is an approach that's served me fine up to this point in my career, why not in the future as well? 🤞