Introduction to Aggregated Apiserver

Overview

The main reason for extending Kubernetes’ core APIs would be those APIs not meeting our business requirements any more. Kubernetes has introduced two primary extension mechanisms to extend Kubernetes APIs— CustomResourceDefinition (CRD) and the aggregated apiserver. They both provide RESTful APIs and work with kubectl. We can use kubectl get to get or list the resources, just like how we list Pods.

What is the aggregated apiserver?

The aggregated apiserver (also known as the custom apiserver) is a standalone HTTP server that provides REST APIs. However, it works totally differently from CRD. In the aggregated apiserver, we can develop fully featured Kubernetes APIs. In straightforward terms, the aggregated apiserver works like another kube-apiserver. It cascades to the kube-apiserver to provide additional APIs that are not offered by the core Kubernetes APIs. This is why we use the word aggregated to describe it. The aggregated apiserver sits behind the kube-apiserver. This kind of arrangement is also called API Aggregation (AA).

We can dynamically register our aggregated apiservers to the kube-apiserver by creating an APIService object below. The kube-apiserver determines where a request should be sent and knows how to contact it. With this object, ...