Kubernetes Architecture Explained (Without the Confusion)

BHARAT PRAKASH INANI — Thu, 23 Apr 2026 08:35:08 +0000

If Kubernetes ever felt like a black box, you’re not alone.

Let’s break it down 👇

🧠 The Core Idea

Kubernetes splits responsibilities into two parts:

Control Plane → makes decisions
Worker Nodes → run your apps

That’s it. Everything builds on top of this.

🔷 Control Plane (The Brain)

This is where all decisions happen.

API Server → entry point
etcd → stores cluster state
Scheduler → assigns workloads
Controller → fixes things when they break

You don’t deploy apps here — you control them.

⚙️ Worker Nodes (The Muscle)

This is where your app actually runs.

Each node has:

Kubelet
Kube-proxy
Container runtime
Pods

👉 Pods = smallest deployable unit

🌐 How Traffic Works

Kubernetes networking is actually simple:

User → LoadBalancer → Service → Pod

Services make sure:

Traffic is balanced
Pods are discoverable
Failures don’t break things

☁️ Managed Kubernetes (The Game Changer)

Instead of setting all this up manually, cloud providers handle it for you:

Amazon EKS
Azure Kubernetes Service (AKS)
Google Kubernetes Engine (GKE)

They manage:
✅ Control plane
✅ Scaling
✅ Reliability

You focus on shipping code.

⚖️ Quick Take
EKS → best for AWS-heavy teams
AKS → best for enterprise / Microsoft stack
GKE → easiest + most automated

🧠 Final Thought

Kubernetes isn’t complicated — it’s just layered.

Once you understand:
👉 Control Plane vs Worker Nodes

Everything starts to click.

From Production Issues to Medium Articles – Sharing My DevOps Journey

BHARAT PRAKASH INANI — Tue, 16 Dec 2025 10:06:59 +0000

Hey everyone 👋

I’m sharing practical DevOps, CI/CD, Cloud & real-world engineering learnings on Medium.

No fluff — only things I’ve worked on in real projects.

Happy to connect with fellow techies 🙌