In this blog post, I will try to explain the relation between Prometheus, Heapster, as well as the Kubernetes metrics APIs and conclude with the recommended way how to autoscale workloads on Kubernetes.
This post assumes you have a basic understanding of Kubernetes and monitoring.
Heapster provides metric collection, basic monitoring capabilities and supports multiple data sinks to write the collected metrics to. The code for each sink resides within the Heapster repository. Heapster also enables the use of the Horizontal Pod Autoscaler to autoscale on metrics.
Read more at Frederic Branczyk blog