Kubernetes - Components

Kubernetes is built from several core components that work together to run and manage containerized applications. These components are usually grouped into:

1. Control Plane Components (manage the cluster)
2. Node Components (run workloads)

1️ . Control Plane Components

The control plane is responsible for managing the Kubernetes cluster.

1. kube-apiserver

The API server is the entry point to the cluster.

Everything in Kubernetes goes through it.

Responsibilities

• Exposes the Kubernetes REST API
• Validates requests
• Processes CRUD operations for resources
• Communicates with the cluster datastore

Example operations:

• Creating a Pod
• Scaling deployments
• Updating configurations

Flow example:

 

User → kubectl → API Server → etcd

 

Key points:

• Stateless
• Can be horizontally scaled
• All components talk to it

 

2. etcd

 

etcd is the key-value database of Kubernetes.

 

It stores all cluster state.

 

Stores

 

Pods

 

Deployments

 

ConfigMaps

 

Secrets

 

Node info

 

Service definitions

 

Example:

 

/registry/pods/default/nginx-pod

 

Key features:

 

Distributed

 

Highly consistent

 

Uses Raft consensus algorithm

 

If etcd is lost → cluster state is lost.

 

3. kube-scheduler

The scheduler decides which node should run a pod.

When a Pod is created:

1. API server receives request
2. Pod has no node assigned
3. Scheduler selects best node

Scheduling decisions based on

• CPU / Memory availability
• Node labels
• Taints and tolerations
• Affinity / Anti-affinity
• Resource requests

Example:

 

Pod → Scheduler → Node3 selected

 

Scheduler only assigns the node.

4. kube-controller-manager

Runs controllers that maintain cluster state.

Controllers constantly watch the cluster and make corrections if needed.

Example idea:

 

Desired state != Actual state

Controller fixes it

Major Controllers

Node Controller

• Detects node failures

Replication Controller

• Ensures correct number of pods

Deployment Controller

• Manages rolling updates

Job Controller

• Handles batch jobs

Example:

If 3 replicas required but only 2 running:

 

Controller creates new pod

5. cloud-controller-manager (optional)

Used when Kubernetes runs on cloud providers.

Manages cloud resources like:

• Load balancers
• Storage volumes
• Node lifecycle
• Routes

Examples of providers:

• AWS
• Azure
• GCP

2️⃣ Node Components

These run on every worker node and are responsible for running containers.

 

1. kubelet

The kubelet is the node agent.

It ensures containers are running as expected.

Responsibilities

• Communicates with API server
• Watches PodSpecs
• Starts containers via runtime
• Reports node status

Example workflow:

 

API Server → Pod assigned to node

       ↓

Kubelet reads PodSpec

       ↓

Starts container

 

2. Container Runtime

The container runtime runs containers.

Examples:

• containerd
• CRI-O
• Docker (older clusters)

Responsibilities:

• Pull images
• Start containers
• Stop containers
• Manage container lifecycle

Example:

 

Kubelet → containerd → start container

 

3. kube-proxy

kube-proxy manages networking rules for services.

It allows pods to communicate using Services.

Responsibilities

• Maintains iptables/IPVS rules
• Implements service load balancing
• Routes traffic to correct pods

Example:

 

Service IP → kube-proxy → Pod1 / Pod2 / Pod3

 

kube-proxy provides internal load balancing.

 

3️⃣ Additional Important Kubernetes Objects

These are not core binaries but are important resources.

Pod

Smallest deployable unit.

Contains:

• One or more containers
• Shared network
• Shared storage

Example:

 

Pod

 ├── Container A

 └── Container B

 

Deployment

Manages stateless applications.

Features:

• Rolling updates
• Rollbacks
• Replica management

Example:

 

Deployment

   ↓

ReplicaSet

   ↓

Pods

 

Service

 

Provides stable networking for Pods.

 

Types:

 

ClusterIP

 

NodePort

 

LoadBalancer

 

Headless

 

Example:

 

Client → Service → Pods

 

ConfigMap

Stores configuration data.

Example:

• Environment variables
• Application configs

 

 

Secret

Stores sensitive data.

Examples:

• Passwords
• API keys
• TLS certificates

 

4️⃣ Kubernetes Architecture Overview

 

 

 

 

 

🧠 Simple Way to Remember

 

Component	Role
API Server	Entry point
etcd	Cluster database
Scheduler	Chooses node
Controller Manager	Maintains desired state
kubelet	Runs pods on node
Container Runtime	Runs containers
kube-proxy	Handles service networking

 

 

✅ One-line summary

Kubernetes works by storing desired state in etcd, exposing it through the API server, scheduling pods with the scheduler, enforcing state with controllers, and running workloads through kubelet and container runtime.