Years ago we lived in a world of monoliths as core of system. Then came times when we started having hundreds and thousands of microservices. In microservices architecture, diagnosing and debugging could be extremely challenging during outages. It is a tedious process to understand how various parts are connected with each other to serve a request. Having monitoring dashboards help, but they alone don’t serve the purpose of debugging in production. How do engineers who are not devops / SREs deal with it? A set of strategy is required to address outages in a timely manner. This talk tells about how bigger companies like Uber use observability and development principles to untangle complex production code and to keep system running even when thousands of developers work with ample of individual microservices.