Kubernetes Architecture

Kubernetes Architecture consist of physical or virtual machine which we called them as nodes. Nodes are cluster services that actually do work. Each nodes have multiple pods or resources running on it.

It consist of two types of nodes :

1. Worker Nodes

2. Master Node

Worker node components

Each worker node should have three processes installed on it.

1. Container runtime

2. Kubelet

3. Kube-proxy

Container Runtime

Used to run container services in node for E.g.: Docker,containerd.

Kubelet

• An agent that runs on each node in the cluster.

• Its a process that actually schedules those pod and containers inside it.

• Kubelet interact with both container and node.

• It Assigns resources after taking it from node like (CPU memory etc.)

• it takes configuration and running the pods.

• Also It makes sure that containers are running in a Pod.

Kube-proxy

• Makes sure your pods and services can communicate with other pods and services in nodes and control plane.

Master Node components

Each master node should have atleast four processes installed on it.

1. Etcd

2. Kube-Scheduler

3. Kube-API Server

4. Kube Controller Manager

Etcd

It is a distributed reliable key-value store used by Kubernetes to store all data used to manage the cluster

Kube-Scheduler

It is responsible for distributing work (newly created pods) across multiple nodes.

Kube-API server

• Exposes the k8s API to create and manage k8s objects like pods, services, ingress. These objects have theirendpoints .

• K8s API has REST interface.

• Kubectl and kubeadm are CLI tools to communicate with k8s API via http request. To see all the k8 object and their api version type below.

 kubectl api-resources

Kube-controller manager

Loops that run continually and check the status of cluster to make sure things are running properly.

For e.g.

Node controller manager continuously monitoring nodes and if any goes down it replaces it.

Replication controller manager helps run multiple instances of pod in cluster to provide high availability/load balancing/scaling. It can also help if only pod is running and it fails then replication controller help by instantly running another replica of pod. So in short it helps run specified number of pods all the time.