4.Five Words on a Map

The Problem in Real Life

A Wall of Jargon vs. Five Words on a Map

Clusters, Nodes, Pods, Deployments and Services

• Cluster — the whole port: a cluster is the whole Kubernetes system: a group of machines working together, plus the control plane — the harbour master's office — which stores the desired state and makes all the decisions. You talk to the cluster, not to individual machines.
• Node — a ship: a node is one machine in the cluster (usually a cloud VM, Act 20) that runs containers. A cluster might have 3 nodes on a quiet day and 15 on Sale Day. If a node dies, its work moves to other nodes.
• Pod — a container (or a few) on a ship: a pod is the smallest thing Kubernetes runs: usually one container, sometimes a few that must always live together and share a network address. Each pod gets its own IP address. Pods are temporary: they're created, killed and replaced all the time, and a replacement gets a new address.
• Deployment — the shipping order: a Deployment describes the desired state for a set of identical pods: which image, how many copies (replicas), how much CPU and memory. It keeps that number running (self-healing), and when you change the image version, it performs a rolling update — and can roll back (Act 19). You almost never create pods by hand; you create a Deployment and it manages the pods.
• Service — the port office with a fixed address: since pods keep changing addresses, other parts of the system need something stable. A Service gives a set of pods one fixed name and address and spreads requests across whichever pods are currently healthy — a built-in load balancer (Act 14). The email worker doesn't need to know where the Redis pods are today; it just connects to the Service named redis.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web
spec:
  replicas: 4                       # always keep 4 copies running
  template:
    spec:
      containers:
        - name: web
          image: registry.blueticket.example/blueticket-web:3.1.0
          ports:
            - containerPort: 3000
          resources:
            limits: { cpu: "1", memory: "1Gi" }