Henri leads the engineering efforts related to Prometheus at Sysdig. He fell in the observability cauldron many years ago, before it even had a name, while observing packets at Riverbed, and has since worked in web monitoring, streaming log analytics, and now security and monitoring at Sysdig.
Prometheus has become the go-to system for monitoring cloud-native infrastructures such as Kubernetes, with lots of integration points and options.
In this talk, we will look at all the relevant pieces for setting up Prometheus monitoring of your Kubernetes cluster - primarily from a user's perspective but also peeking under the hood to see how things work.
On the Prometheus side, we'll cover the different ways that service discovery and relabelling can be used, and how they make Prometheus so flexible.
On the Kubernetes side, we'll look at the alphabet soup of available metrics exporters (kube-state metrics, cAdvisor, node exporter, and friends).
Finally, we'll take a step back and consider how Prometheus' somewhat unique design and feature set fits into the broader landscape of monitoring systems, and how it is especially suited to cloud-native environments like Kubernetes.