I enjoyed this concrete look at how Slack moved to fully automated deployments, leaving it all to the robots. Building out all the tooling you need to do continuous delivery is a challenge, but far from an insurmountable one. I suspect most organizations have a turbo-nerd ready and eager to program it all up. The bigger issue (as always?) is the non-technical: people are scared something will break when no one is looking. This blog post discusses one way to effectively monitor for those sorts of situations and bubble them up: z-scores. This approach to monitoring is seems useful for any alerting an ops team gets up to.
