In previous blog posts, we have talked about the basics of autoscaling, autoscaling using custom metrics and Prometheus Operator which covers various aspects of monitoring in Kubernetes. One thing we haven’t talked so much about is the high availability (HA) of Prometheus in the cluster. What if the single Prometheus instance goes down? Or if a single instance can not handle metrics for the whole cluster and you need to scale horizontally?In this post, we will use Thanos to make Prometheus highly available (HA) and scalable. Sounds exciting? Let’s get started!

#prometheus #thanos #monitoring #observability