SLO-Based Observability For All Kubernetes Cluster Components

In this talk, Nadine and Matthias will give an introduction to Pyrra, a project that aims to make Service Level Objectives (SLOs) with Prometheus manageable, accessible, and easy to use for everyone. Nadine will talk about the project approach and findings for creating an easy-to-use observability tool. Matthias will then walk the audience through setting up a Pyrra instance on Kubernetes and how to connect it with either Prometheus or Thanos. After a successful deployment every component of the cluster will get an SLO, starting with etcd, the Kubernetes API server and kubelet, CoreDNS, and at the end Prometheus and Pyrra itself. In the end, a demo will showcase an outage in the cluster and what the alerting will look like, discussing how the lives of on-call engineers have been improved.