Kubernetes Deep Dive: Architecture, Pods, and Beyond

• API Server — The front door for all REST commands; the CLI (kubectl) talks to it.
• etcd — A consistent and highly-available key-value store for all cluster data.
• Scheduler — Assigns newly created pods to appropriate nodes.
• Controller Manager — Runs controller processes that regulate the cluster state.

Worker nodes are the machines (virtual or physical) that run the actual application pods. Each node runs:

• Kubelet — An agent that ensures containers are running in a Pod as expected.
• Container Runtime — The software responsible for running containers (e.g., containerd, CRI-O).
• Kube-proxy — Maintains network rules for pod communication.

Pods: The Smallest Deployable Unit

You never deploy a container directly in Kubernetes — you deploy a Pod. A Pod is one or more containers that always co-locate and co-schedule.

apiVersion: v1
kind: Pod
metadata:
  name: web-pod
  labels:
    app: web
spec:
  containers:
  - name: nginx
    image: nginx:1.27
    ports:
    - containerPort: 80
    resources:
      requests:
        memory: "64Mi"
        cpu: "250m"
      limits:
        memory: "128Mi"
        cpu: "500m"

In practice, you rarely create Pods directly. Instead, you use higher-level abstractions like Deployments.

Deployments and Services

A Deployment manages replicated Pods and handles rolling updates:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web-deployment
spec:
  replicas: 3
  selector:
    matchLabels:
      app: web
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
      - name: nginx
        image: nginx:1.27
        ports:
        - containerPort: 80

A Service provides a stable network endpoint for your Pods:

apiVersion: v1
kind: Service
metadata:
  name: web-service
spec:
  selector:
    app: web
  ports:
  - protocol: TCP
    port: 80
    targetPort: 80
  type: ClusterIP

Namespaces and Resource Management

Namespaces isolate resources within a cluster. Create environments like staging and production to prevent naming conflicts and enforce resource quotas:

apiVersion: v1
kind: Namespace
metadata:
  name: production
---
apiVersion: v1
kind: ResourceQuota
metadata:
  name: prod-quota
  namespace: production
spec:
  hard:
    pods: "20"
    requests.cpu: "4"
    requests.memory: "8Gi"
    limits.cpu: "8"
    limits.memory: "16Gi"

Conclusion

Kubernetes has a steep learning curve, but its abstractions — Pods, Deployments, Services, and Namespaces — form a coherent model for managing complex applications at scale. Start small with a local cluster using kind or minikube, deploy a simple app, and gradually explore more advanced concepts like Ingress, ConfigMaps, and Helm charts.