Being able to observe the state of a running application is the key to understanding a system’s behaviour and being more efficient when failure happens. This talk focuses on why observability should be an integral part of system design and on the techniques to build a clearer picture of distributed systems in production.