DevOps

Defined Kubernetes in Gaurav Arora Words.

What is Kubernetes ?

  • Kubernetes is borned by Google
  • It is Open Source platform
  • It is cncf product
  • It overcomes challenges like scalability challenges
  • We can run 1000 computers via Kubernetes
  • It can help running multiple container with same port.
  • It will resolve auto healing in case of any failure
  • It solve networking challenge also
  • It will help managing multiple docker host

Why Kubernetes ?

  • It helps to run multiple docker container
  • it worked in desired state
  • Speed of deployment is faster and improved
  • It recover quickly
  • It helps in hiding complexity in Cluster
  • it helps in doping backup automatically
  • It is Secret and Configuration management
  • It is self healing and quickly recover in few seconds
  • It apparently enhances productivity
  • It lower overhead as everything will be managed by Kubernetes itself.
  • It increased portability feature.
  • It improved security
  • It is immutable means you will not get same pod

How Kubernetes works ?

Kubernetes basically consists of Cluster Store , Scheduler , Controller and apiserver.

Apiserver manages everything front end to the control plane , It exposes Restful API’s and consumes JSON. Can do CRUD operations and updated etcd also.

Cluster Store is persistent storage of K8 and uses etcd for distributed , consistent and watchable . It basically stored everything on Cluster and basically it is the source of truth.

Controller is managing everything as of node controller , endpoints controller and namespace controller.

Scheduler keeps watching apiserver for new Pods . After it , assigns work to nodes , contraints and resources.

What is POD ?

  • Pod is atomic Unit of Scheduling
  • It is running in worker node
  • It may be in single container or multiple container
  • It can only be instantiated by Kubelet
  • It Got IP addresses from kube-proxy
  • it is your unit of scheduling
  • If we have 2 Pod in same container , it is known as Loose coupling
  • If we have multiple Pod in same container , known as Tight coupling
  • POd lifecycle starts from Pending –> Running —> Succedded/ failed
  • It totally depend on Container , if it is running it should run otherwise stopped.