Kubernetes Basics

What is Kubernetes

• Container orchestration platform
• Handles:
  • Deployment
  • Scaling
  • Networking
  • Self-healing
• Built by Google, now maintained by the CNCF
• Think of it as the "control tower" for container ships (apps)

Core Terminology

• Pod: One or more containers, smallest deployable unit
• Node: A worker machine (VM or physical) that runs pods
• Cluster: Group of nodes managed together
• Deployment: Desired state management for pods
• Service: Abstracts and exposes a group of pods
• Namespace: Isolated environments within a cluster

High-Level Architecture

• API Server: Front door to the cluster
• etcd: Stores all cluster configs
• Scheduler: Places pods on nodes
• Controller Manager: Keeps desired state

Control Plane vs Worker Nodes

Control Plane

• API Server
• Scheduler
• Controller Manager
• etcd (database)

Worker Nodes

• Kubelet (pod manager)
• Kube Proxy (networking)
• Container runtime (e.g., containerd)
• Runs the actual Pods

Deployment Methods

• Minikube – Local development
• kubeadm – Manual cluster setup
• Managed Services:
  • GKE (Google)
  • EKS (AWS)
  • AKS (Azure)
• Helm – Kubernetes package manager

Choose based on scale and team size

Declarative vs Imperative

Imperative:

kubectl run nginx --image=nginx

Declarative:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx
spec:
  replicas: 2
  template:
    spec:
      containers:
      - name: nginx
        image: nginx

Prefer declarative (YAML) for production & GitOps

Ecosystem Tools

• Helm: App packaging & deployment
• Prometheus + Grafana: Monitoring
• ArgoCD: GitOps for K8s
• Istio / Linkerd: Service mesh
• K9s: Terminal UI for K8s clusters

Summary & Takeaways

• Key concepts: Pods, Nodes, Services, Deployments
• Understand Control Plane vs Worker Nodes
• Choose the right deployment method
• Embrace declarative infrastructure

Questions / Demo?