True Observability Needs High-Cardinality

The growing complexity of architectures and applications today makes it impossible to cover all the things that may go wrong. These unknown-unknowns are an inevitable component of distributed systems in production and cannot be solved with generic system metrics or by searching through billions of unstructured logs. Being able to explore a running application is key to understanding emergent behaviours and needs to be done with as much preserved context as possible.  

In this session, I’ll cover how high-cardinality observability unleashes the exploration and debugging power required to properly and quickly understand the reality of our production systems.