SREcon16 Europe

This talk will debunk some common beliefs that in order to solve more [advanced] monitoring use cases and get more complete alerting coverage, we need complex, often math-oriented solutions such as machine learning and stream processing.

Instead we will set a clear context and pro's/cons for such approaches, and zoom in on how we can get dramatically better alerting, as well as make our lives a lot easier by using familiar concepts understandable to everyone such as basic logic, basic math and metric metadata, even for solving complicated alerting problems.

We will also see how we can optimize the overall experience of adjusting and maintaining alerting rules over time by focusing on the concept of an alerting IDE, exemplified by bosun. The talk will present techniques and concrete examples of how to do advanced alerting scenarios using these principles.