With a distributed production container deployment with thousands of containers they need to discover and chat with each other. With Docker, creating a highly decoupled microservice architecture is easier than ever, making service discovery and routing a hot topic. If not executed properly, such a system can easily become a reliability nightmare with your service discovery cluster becoming a single point of failure or a single unreliable service compromising the performance or availability of the entire system. With experience from a Docker deployment of thousands of containers, this talk covers the patterns for success to engineer a distributed and resilient routing and discovery system at scale.