To achieve this, it is necessary to closely monitor and observe the metrics and data sets associated with the functionality of computer systems. And that’s precisely what observability is all about: the ability to infer the internal conditions of a system from its external outputs. In control theory, this observability goes hand in hand with controllability (monitoring), namely the ability to control the internal states of a system based on internal inputs. In other words, observability consists in accessing the data of a system that you want to follow, then collecting, viewing and analyzing it. Because once the data has been collected using a tool, it is necessary to perform a manual or – preferably – automated analysis to draw conclusions and take any necessary corrective measures.
Observability is therefore a new approach aimed at rethinking the models and methods of monitoring information systems in order to adapt them to current realities, in particular the cloud and virtualization. In short, it is a question of monitoring any event likely to impede the proper functioning of an organization, what some call the “unknown unknowns” of the IT environment: these facts of which we are not aware, that we do not understand or do not monitor.