Advanced Kubernetes: 1 Core Concepts

Explore the internal architecture of a Kubernetes cluster, and other core concepts needed for an advanced understanding of Kubernetes.

Introduction

• Exploring the Kubernetes architecture
• What you should know

1. Kubernetes Architecture Overview

• The Kubernetes cluster architecture from 40,000 feet
• Who interacts with the cluster?
• Storing resources in the database
• Making things happen
• Achieving high availability
• Scaling with load
• Bringing it together: Control plane load

2. Etcd and API server: The Core of the System

• Some terminology: Kernels, user space, and distributions
• Looking at resources in etcd
• Manually querying the Cluster API

3. Kubernetes API Server: The Life of a Request

• Validating identity with authentication
• Checking permissions with authorization
• Intercepting requests with admission controllers
• Extending admission control with webhooks
• Intelligent persistence and retrieval of resources
• Bringing it together: apply, diff, and dry-run

4. Reconciliation: The Engine of a Declarative System

• Putting the world right with reconciliation
• Everything's a controller
• Example controller: Deployment
• Who controls controllers? The controller-manager
• The many other loops of the controller-manager
• Controlling pods' placement: The scheduler
• Bringing it together: Scheduler mechanics

5. Data-Plane Reconciliation

• What's contained in a pod?
• Controlling pods with the kubelet
• kube-proxy: The iptables controller
• Weather machine: The cloud-controller-manager
• Discovering services with cluster DNS
• Bringing it together: The operator pattern

Conclusion

• Next steps