K8s Architecture

by KAUSTUBH SHARMA on July 26, 2021, 1:24 a.m. with 540 views

Learn about Kubernetes Architecture in detail from the below blog.

 

 

 

 

 

 

CLUSTER

• Made of multiple nodes.
• Interacts with both conatiner as well as node(machine)
• Reposnible for running/creating pod within a container & assigning resources like CPU,RAM etc. to container from that node

 

 

• One of the main components of kubernetes architecture are its worker servers or nodes
• Each node will has multiple pods on it.
• Nodes are the cluster servers that actually do the work that's why also called worker nodes.
• 3 process must be installed on each node that are used to schedule and manage those pods.

 

3 PROCESSESS MUST BE ON EACH WORKER NODES

• Container Runtime like Docker.
• Kubelet
• Kube proxy

 

KUBELET

• Process of K8s itself.
• Interacts with both conatiner as well as node(machine)
• Reposnible for running/creating pod within a container & assigning resources like CPU,RAM etc. to container from that node

 

KUBE proxy

• Responsible for forwarding requests from services to pods.
• Has intelligent forwarding logic inside that makes sure that the communication also works in a performant way with low overhead like sending DB pod request to same node app pod.

 

MASTER NODES

• Schedule pods, restart/reschedule pod, monitor, join a new node 7 other components.
• Need less resources(CPU/RAM etc.) than worker nodes(do actually work).

 

4 PROCESSESS MUST BE ON MASTER NODES

• API Server
• Scheduler
• Controller Manager
• etcd

 

API Server

• Acts like Cluster gateway, as it get initial request of any updates into the cluster or even the queries from the cluster
• Acts as gatekeeper for authentication means allows only authenticated and authorized requests get through to the cluster.
• Only entry Point into cluster.

 

Scheduler

• Intelligently decides on which worker node, next component/pod will be scheduled based on resources need to complete your request.
• Remember Scheduler only decides, while Kubelet executes the request.

 

Controller Manager

• Detects cluster state changes like any pod dies & make scheduler a request to reschedule those dead pods.

 

etcd

• Key value store of a cluster state
• Cluster changes stores in this key value store
• Cluster brain-all other process work based on this data
• Actual app data is not stored in etcd

 

In practice, the Kubernetes cluster is usually made up of multiple masters where each master node runs its master processes where of course the api server is load balanced and the etcd store forms a distributed storage across all the master nodes.

 

Learn more this through Video on -YouTube