Kubernetes an Overview

Kubernetes an Overview

Kubernetes is an open-source container-orchestration system for automating application deployment, scaling, and management. It was originally designed by Google, and is now maintained by the Cloud Native Computing Foundation

Kubernetes is the de facto standard for running containerized applications.

Kubernetes (K8s) is an open-source system for automating deployment, scaling, and management of containerized applications.

Kubernetes makes it easy to deploy and run containerized applications. Kubernetes is simple to use.

Kubernetes is complex to understand because it provides a huge set of options to make your deployment easier.

Aptly named, Kubernetes is a pilot (or) helmsman that helps you to sail the container world. Kubernetes is a portable and extensible system built by the community for the community.

As Kelsey, correctly quotes

Kubernetes does the things that the very best system administrator would do automation, failover, centralized logging, monitoring. It takes what we’ve learned in the DevOps community and makes it default, out of the box.

In order to work with Kubernetes, it is very important to understand

  • How Kubernetes works?
  • How Kubernetes is architected?
  • What are the various components in Kubernetes?

Let us start hacking on Kubernetes.

How does Kubernetes work?

The Kubernetes run in a highly available cluster mode. Each Kubernetes cluster consists of one or more master node and a few worker nodes.

Master Node

The master node consists of an API server, Scheduler, Controllers, etcd. This node is called the control plane of Kubernetes. This control plane is the brain of Kubernetes.

That is the control plane is responsible for all the actions inside Kubernetes.

Via the API server, we can instruct the Kubernetes or get information from the Kubernetes.

The Scheduler is responsible for scheduling the pods.

The controllers are responsible for running the resource controllers.

The etcd is a storage for the Kubernetes. It is key-value storage.

Worker Node

The worker nodes have a Kubelet and proxy.

The Kubelets are the actual workhorse and the Kube-proxy handles the networking.


We provide the yaml file to the Kubernetes cluster through kubectl apply command.

The apply command calls the API server, which will send the information to the controller and simultaneously stores the information to the etcd.

The etcd then replicate this information across multiple nodes to survive any node failure.

The controller will check whether the given state matches the desired state. If it is not it initiates the pod deployment, by sending the information to the scheduler

The checks are called as the reconciliation loop that runs inside the Kubernetes. The job of this loop is to validate whether the state requested is maintained correctly. If the expected state and actual states mismatch this loop will do the necessary actions to convert the actual state into the expected state.

The scheduler has a queue inside. Once the message is received in the queue.

The scheduler will then invoke the kubelet to do the intended action such as deploying the container.

This is a 10000 feet bird view of how Kubernetes does the deployment.

There are various components inside the Kubernetes. Let us take a look at what are they and how are they useful.

Components of Kubernetes


In general terms, pods are nothing but a group of dolphins or whales.

Similarly, in Kubernetes world, pods are a group of containers living together. A pod may have one or more containers in it.

The pod is the smallest unit of deployment in Kubernetes. Usually, the containers that cannot live outside the scope of another container are grouped to form a pod.

This is how you define a pod in Kubernetes.

apiVersion: v1
kind: Pod
 name: myapp-pod
   app: myapp
 - name: myapp-container
   image: busybox
   command: ['sh', '-c', 'echo Hello Kubernetes! && sleep 3600']
  • apiVersion denotes the Kubernetes cluster which version of API to use when parsing and executing this file.
  • kind defines what is the kind of Kubernetes object, this file will refer to.
  • metadata includes all the necessary metadata to identify the Pod.
  • spec includes the container information.


While pods are the unit of deployment. For an application to work, it needs one or more pods. Kubernetes considers this entire set as deployment.

Thus deployment is recorded information about pods. Kubernetes uses this deployment information to manage and monitor the applications that are deployed in them.

The below file is the sample deployment file that tells the Kubernetes to create a deployment of nginx using the nginx:1.7.9 container.

apiVersion: apps/v1
kind: Deployment
 name: nginx-deployment
   app: nginx
 replicas: 3
     app: nginx
       app: nginx
     - name: nginx
       image: nginx:1.7.9
       - containerPort: 80


While deployment tells the Kubernetes what containers are needed for your application and how many replicas to run. The replica sets are the ones that ensure those replicas are up and running.

ReplicaSet is responsible for managing and monitoring the replicas.


Often times we will need to have persistent storage or permanent network identifiers or ordered deployment, scaling, and update. During those times we will use StatefulSets.

You can define the StatefulSet like below:

apiVersion: apps/v1
kind: StatefulSet
 name: web
     app: nginx # has to match .spec.template.metadata.labels
 serviceName: "nginx"
 replicas: 3 # by default is 1
       app: nginx # has to match .spec.selector.matchLabels
     terminationGracePeriodSeconds: 10
     - name: nginx
       image: k8s.gcr.io/nginx-slim:0.8
       - containerPort: 80
         name: web
       - name: www
         mountPath: /usr/share/nginx/html
 - metadata:
     name: www
     accessModes: [ "ReadWriteOnce" ]
     storageClassName: "my-storage-class"
         storage: 1Gi

We mounted the volume and also claimed the volume storage.


Sometimes you need to run a pod on every node of your Kubernetes cluster. For example, if you are collecting metrics from every node, then we will need to schedule some pods on every node that collects the metrics. We can use DaemonSet for those nodes.


The deployments define the actual state of the application running on the containers. Users will need to access the application or you might need to connect to the container to debug it. Services will help you.

The services are the Kubernetes object that provides access to the containers from the external world or between themselves.

We can define the service like below:

apiVersion: v1
kind: Service
 name: my-service
   app: MyApp
 - protocol: TCP
   port: 80
   targetPort: 9376

The above service maps incoming connections on port 80 to the targetPort 9376.

You can consider the services as the load balancer, proxy or traffic router in the world of Kubernetes.


This is the most important element of Kubernetes. The pods running should be exposed to the network. The containers that are running inside the pods should communicate between themselves and also to the external world.

While service provides a way to connect to the pods, networking determines how to expose these services.

In Kubernetes we can expose the service through the following ways:

  • Load Balancer
  • The Load Balancer provides an external IP through which we can access the pods running inside.
  • The Kubernetes will start the services and then asynchronously starts a load-balancer.

  • Node Port
  • Each of the services will have a dynamically assigned port.
  • We can access the services using the Kubernetes master IP.

  • Ingress
  • Each of the services will have a separate address.
  • These services are then accessed by an ingress controller.
  • The ingress controller is not a public IP or external IP.


Often for the applications, we need to provide passwords, tokens, etc., Kubernetes provides secrets object to store and manage the sensitive information. We can create a secret like below:

apiVersion: v1
kind: Secret
 name: mysecret
type: Opaque
 config.yaml: |-
   apiUrl: "https://my.api.com/api/v1"
   username: {{username}}
   password: {{password}}

Best practices

While Kubernetes is an ocean and whatever we have seen is just a drop in it. Since Kubernetes supports a wide range of applications and options, there are various different options and features available.

Few best practices to follow while working with Kubernetes are:

Make smaller YAML

The yaml files are the heart of Kubernetes configuration.

We can define multiple Kubernetes configurations in a single yaml. While yaml reduces the boilerplate when compared with JSON. But still yaml files are space-sensitive and error-prone.

So always try to minimize the size of yaml files.

For every service, deployment, secrets, and other Kubernetes objects define them in a separate yaml file.

Split your yaml files into smaller files.

The single responsibility principle applies here.

Smaller and Fast boot time for images

Kubernetes automatically restarts the pods when there is a crash or upgrade or increased usage. It is important to have a faster boot time for the images. In order to have a faster boot time, we need to have smaller images.

Alpine images are your friends. Use the Alpine images as the base and then add in components or libraries to the images only when they are absolutely necessary.

Always remember to have smaller image sizes. Use builder pattern to create the images from Alpine images.

Healthy - Zombie Process

Docker containers will terminate only when all the processes running inside the container are terminated. The Docker containers will return healthy status even when one of the processes is killed. This creates a Healthy-Zombie process.

Try to have a single process inside the container. If running a single process is not possible then try to have a mechanism to figure out whether all the required processes are running.

Clean up unused resources

In the container world, it is quite common to have unused resources occupying the memory. It is important to ensure the resources are properly cleaned.

Think about Requests & Limits

Ensure that requests and limits are properly specified for all the containers.

       memory: "100Mi"
       cpu: "100m"
       memory: "200Mi"
       cpu: "500m"

The requests are the limits that the container is guaranteed to get. The limits are is the maximum or minimum resource a container is allowed to use.

Each container in the pod can request and limit their resources.

RED / USE pattern

Monitor and manage your services using RED pattern.

  • Requests
  • Errors
  • Duration

Track the requests, errors in the response and the duration to receive the response. Based on this information, tweak your service to receive optimum performance.

For the resources, use the USE pattern.

  • Utilization
  • Saturation
  • Errors

Monitor the resource utilization and how much the resources are saturated and what are the errors. Based on this information, tweak your resources to optimize resource allocation.

Hopefully, this might have given you a brief overview of Kubernetes. Head over kubernetes.io for more information on Kubernetes.

Thanks for reading

If you liked this post, share it with all of your programming buddies!

Follow me on Facebook | Twitter

Further reading about Kubernetes

Docker and Kubernetes: The Complete Guide

Learn DevOps: The Complete Kubernetes Course

Docker and Kubernetes: The Complete Guide

Kubernetes Certification Course with Practice Tests

An illustrated guide to Kubernetes Networking

An Introduction to the Kubernetes DNS Service

Kubernetes Deployment Tutorial For Beginners

Kubernetes Tutorial - Step by Step Introduction to Basic Concepts

kubernetes devops docker serverless

Bootstrap 5 Complete Course with Examples

Bootstrap 5 Tutorial - Bootstrap 5 Crash Course for Beginners

Nest.JS Tutorial for Beginners

Hello Vue 3: A First Look at Vue 3 and the Composition API

Building a simple Applications with Vue 3

Deno Crash Course: Explore Deno and Create a full REST API with Deno

How to Build a Real-time Chat App with Deno and WebSockets

Convert HTML to Markdown Online

HTML entity encoder decoder Online

DevOps Tutorial - Docker, Kubernetes, and Azure DevOps

In this article, we break down everything you need to know about DevOps, so that you can get started building your own CI/CD pipeline. DevOps Tutorial - Docker, Kubernetes, and Azure DevOps

Kubernetes vs Docker

Get Hands-on experience on Kubernetes and the best comparison of Kubernetes over the DevOps at your place at Kubernetes training

DevOps Event for CI/CD, Serverless and Kubernetes

DevOps Events Includes various tracks CI/CD, Cloud-Native Infrastructure, DevSecOps, Cultural Transformations, and SRE

Docker vs. Kubernetes | Docker vs. Kubernetes Difference

Docker and Kubernetes are two orchestration tools that very popular. Many people have trouble picking one. In this video on Docker vs Kubernetes, we will be comparing these two tools end to end and see which one will suit your needs better.

DevOps and Docker Live Show (Ep 91)

Join me with guest Docker Captain Elton Stoneman to talk about the state of Docker Desktop and Docker Hub. Support this show on Patreon! It's the #1 way to support me interviewing DevOps and container experts, and doing this Live Q&A.