I've always found deploying a cluster management service to be a very annoying overhead to any distributed system. It's often the source of production problems, and requires additional resources that just don't really do anything. Each cloud provider has a queue system (Kinesis for AWS, Event Hub for Azure, etc.) and most production deployments include a Kafka cluster. In this talk, I'll show how we created a "headless" managed cluster using a distributed queue system, which reduced cluster deployment overhead while increasing visibility into what's happening in the cluster.