Deploying into Production: The Need for a RED Light

Testing in production is accepting the reality on the ground. App complexity and scale mean pre-deployment testing becomes harder, slower, and more expensive. Once we accept this reality, we look for other ways to mitigate risk. Machine learning can help here by looking for bad stuff happening which may never have been seen before, and alerting us, usually with some indication of root-cause. Larry walks you through an example of how this approach works from a database of annotated real-world incidents.