We all know logs, metrics and traces, the “three pillars of observability”. We’ve been told that by collecting them we’d gain observability into our systems, right? WRONG!

Observability is NOT logs+metrics+traces. You can diligently collect these signals and still find yourself without the required observability to detect and root-cause during a major outage or incident. We need a paradigm shift. Observability is actually a data analytics problem.

In this inspiring and thought provoking talk, Horovits will introduce the data analytics approach, together with practical measures that will guide you in gaining real observability into your system and in getting the insights you need, when you need them. Horovits will also challenge the “holy pillar trinity” and look into additional observability data you may not have considered, and other conventions you've grown used to.