Modern distributed computer systems have given rise to a seeming unlimited set of ill-defined failure modes. So when something goes wrong, finding the root cause can be tough.
To make sense of this new paradigm, we need to shift mindset, moving from monitoring to observability — our ability to determine the internal states of our systems from their outputs.
