When referring to Data in the cloud context, engineers may often think of them as time series. Time series are a Data format storing variables values along with their timestamps, as for example system or sensor metrics. Data help to identify current execution status, as well as behavior or trends by sysadmins, and are used to build predictive models by data scientists.
While there is a lot of information held in time series, they may not be expressive by themselves. They can either come in too great of a number, size, pace, or need specific transformation to provide end-user, human, readability. Nowadays, Cloud providers propose additional services, which deliver domain oriented information based on time series analytics, but these additional services are for the most part not free and very specialized.
Rating-Operator is an open source Kubernetes application aiming to provide sysadmins and domain-oriented users, such as business developers, with KPIs built from time series metrics. These KPIs can either be infrastructure costs, applicative performance or any other kind of execution and business efficiency indicators.
In this talk, we will introduce Rating-Operator and demonstrate its usage : from defining and pushing a set of pricing rules, to making them available as a new datasource and achieve a public Cloud costs comparison use case.