Orchestration and Kubernetes Basics

Introduction

Running a few containers on one server is easy. Running hundreds across many servers, restarting failures, updating without downtime and scaling with demand is hard to do by hand. An orchestrator automates this. The most widely used one is Kubernetes.

What an orchestrator does

• Schedules containers onto machines with enough resources
• Restarts containers that fail, and replaces machines that die
• Scales the number of copies up and down
• Rolls out new versions gradually, and rolls back if needed
• Connects containers with stable network addresses and load balancing
• Stores configuration and secrets

Core Kubernetes ideas

You describe the desired state in files, usually YAML, and Kubernetes continuously works to make reality match.

• Cluster: a set of machines (nodes) managed together.
• Pod: the smallest unit. One or more containers that run together and share a network.
• Deployment: says "keep three copies of this pod running", and handles rolling updates.
• Service: a stable address and load balancer in front of a set of pods, because pods come and go.
• ConfigMap and Secret: configuration and sensitive values supplied to pods.
• Namespace: a way to separate teams or environments inside a cluster.

A small Deployment:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-app
spec:
  replicas: 3
  selector:
    matchLabels:
      app: my-app
  template:
    metadata:
      labels:
        app: my-app
    spec:
      containers:
        - name: my-app
          image: registry.example.com/team/my-app:3f9a2c1
          ports:
            - containerPort: 3000

Apply it with kubectl apply -f deployment.yaml, and look with kubectl get pods.

Do you need it?

Kubernetes is powerful, and it is also complex to learn, run and secure.

• Compose on one server is often enough for small apps and teams.
• Consider managed platforms that run containers for you, such as managed Kubernetes services or simpler container hosting, so you do not run the cluster yourself.
• Choose Kubernetes when you need many services, scaling across machines, zero-downtime updates or a platform used by many teams.

Do not adopt it just because it is popular. Start simple, and move when you feel real limits.

Security in clusters

All the earlier hardening still applies: non-root, least privilege, scanned images and no secrets in images. Also use access controls, network policies to limit which pods can talk, and keep the cluster itself updated.

Where to go next

• Practise with a local tool that runs a small cluster on your laptop.
• Learn kubectl basics, then Helm for packaging.
• Learn monitoring and logging for clusters.

Try it yourself

Take the application you containerised, and sketch which Kubernetes objects it would need: how many pods, what service in front, what config and secrets. Then decide honestly whether Compose would be enough for it today.

Summary

• Orchestrators schedule, restart, scale and update containers across machines.
• Kubernetes uses pods, deployments and services, driven by desired state files.
• Choose it for real scale and complexity, and prefer simpler options until then.