Getting Started with Kubernetes — A Practical Guide

Why Kubernetes?

After running containerized workloads across multiple production environments — from self-managed clusters to AWS EKS — I can confidently say Kubernetes is the right tool for serious container orchestration. It automates deployment, scaling, and operations so you can focus on delivering value, not babysitting infrastructure.

Prerequisites

Docker installed locally
kubectl CLI
A local cluster via minikube or kind
Basic understanding of containers

Your First Deployment

kubectl create deployment hello-app --image=nginx
kubectl expose deployment hello-app --type=NodePort --port=80
kubectl get services

This creates a simple nginx deployment and exposes it via a NodePort service.

Key Concepts to Understand

Pods

The smallest deployable unit. A pod wraps one or more containers that share network and storage.

Deployments

Manage the desired state of your pods — rolling updates, rollbacks, and replica scaling.

Services

Abstract network access to a set of pods. Types include ClusterIP, NodePort, and LoadBalancer.

Cost Optimization with Karpenter

One thing I’ve learned running EKS in production: node provisioning matters for cost. Karpenter is a game-changer — it provisions exactly the right nodes for your workloads, eliminating over-provisioning.

apiVersion: karpenter.sh/v1alpha5
kind: Provisioner
metadata:
  name: default
spec:
  requirements:
    - key: karpenter.sh/capacity-type
      operator: In
      values: ["spot", "on-demand"]
  limits:
    resources:
      cpu: 1000
  ttlSecondsAfterEmpty: 30

Event-Driven Scaling with KEDA

For workloads that need to scale based on external metrics (queues, topics, etc.), KEDA extends Kubernetes autoscaling beyond CPU/memory:

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: my-scaledobject
spec:
  scaleTargetRef:
    name: my-deployment
  triggers:
    - type: aws-sqs-queue
      metadata:
        queueURL: https://sqs.eu-west-1.amazonaws.com/123/my-queue
        queueLength: "5"

Helm — Package Management for Kubernetes

Don’t manage raw YAML manifests at scale. Use Helm:

helm repo add stable https://charts.helm.sh/stable
helm install my-release stable/nginx-ingress

Next Steps

Once you’re comfortable with the basics, explore:

Karpenter for intelligent node provisioning
KEDA for event-driven autoscaling
ArgoCD for GitOps-based deployments
Ingress controllers for HTTP routing

Kubernetes has a steep learning curve, but the investment pays off quickly when managing production workloads at scale.

---