How does RabbitMQ handle network latency? What about a clean network partition? And a partial network partition, or partial switch failures?
We have at our disposal a wide variety of tooling for the Kubernetes infrastructure that will let us make new discoveries about the behaviour of RabbitMQ.