Introduction

Containers solved packaging. A Docker image runs the same way on a laptop, a test server and production. But a container on its own is still just a process on one machine. If the machine dies, the container dies. If traffic doubles, nobody starts more copies. If you push a new image, someone has to stop the old containers and start the new ones, carefully, on every server.

Kubernetes takes over that work. You describe the state you want ("three copies of this image, reachable on port 80, using this configuration") and Kubernetes keeps the cluster in that state: it places containers on machines with room, replaces anything that fails, spreads traffic across healthy copies, and changes versions step by step when you ask it to.

What Kubernetes is

Kubernetes (often written K8s) is an open source container orchestration platform. Google open sourced it in 2014, based on more than a decade of running its own internal system, Borg. Today it is maintained by the Cloud Native Computing Foundation and is the standard way to run containers in production on every major cloud and in private data centres.

It is not a replacement for Docker. Docker (or any OCI tool) builds the images. Kubernetes runs them across a cluster of machines and keeps them running.

Desired state: the one idea behind everything

Almost everything in Kubernetes follows the same loop:

You write YAML (desired state)  ->  kubectl apply  ->  API server stores it in etcd
        ^                                                     |
        |                                                     v
  you change the YAML              controllers compare desired state with actual state
                                   and act until they match (create, delete, replace)

That loop is why a deleted pod comes back, why scaling is a one-line change, and why a rolling update happens without you touching individual containers. Once this idea clicks, every new object in the course is "another kind of desired state with its own controller".

Core vocabulary

Why Amazon EKS

You can run Kubernetes on your laptop (minikube, kind), on your own servers (kubeadm), or as a managed service. This course uses Amazon Elastic Kubernetes Service (EKS):

• AWS runs and patches the control plane across multiple Availability Zones
• Worker nodes are ordinary EC2 instances in a managed node group
• type: LoadBalancer Services create real AWS load balancers, and PersistentVolumes can be real EBS volumes
• eksctl creates the whole setup with one command and deletes it with another

The manifests you write are standard Kubernetes and work on any conformant cluster. Only the cluster setup and a few integrations are EKS specific.

How Kubernetes fits with the rest of the bootcamp

The final Jenkins project of the bootcamp ties these together into one pipeline that ends with a rollout on Kubernetes.

What Kubernetes does not do

Kubernetes is powerful, but it is not magic. It does not build your images, write your Dockerfiles, fix your application's bugs, back up your databases by itself, or make a badly designed app scale. It also adds real operational complexity. For a single small application, a VM with Docker Compose or a simpler service such as Amazon ECS may be the better choice. Kubernetes earns its place when you run many services, deploy often, and need scaling, self-healing and consistent tooling across environments.

Official references

• Kubernetes documentation: Overview
• Kubernetes: Cluster Architecture
• Amazon EKS User Guide
• CNCF: Kubernetes

Written By: Shramik Masti