Introduction to K8s and it's Concept
Concepts: Pods, Nodes, Services, Deployments, ReplicaSets
Learning Outcome
4
Understand how these components work together for scalability
3
Describe how Services enable stable communication
2
Explain how ReplicaSet and Deployment maintain desired state
1
Identify the role of Nodes and Pods in a cluster
Topic Name-Recall(Slide3)
Think of Kubernetes as a Shopping Mall!
Imagine a large shopping mall:
The entire mall = Kubernetes Cluster
Think of Kubernetes as a Shopping Mall!
Each floor = Node
Each shop = Pod
Mall Management = Deployment
Backup shop copies = ReplicaSet
Help Desk Counter = Service
Think of Kubernetes as a Shopping Mall!
If one shop closes suddenly, customers should still be able to buy products
1
If too many customers arrive, more shops should open.
2
If customers don’t know shop locations, they should ask at the help desk
3
This is exactly what Kubernetes does for applications
From Mall Analogy to Real Kubernetes Components
Why do we need these components?
When running applications in containers:
Containers may crash
Traffic may increase suddenly
Servers may fail
Pod IP addresses keep changing
We need scaling and load balancing
From Mall Analogy to Real Kubernetes Components
Without proper management:
Users see errors
Apps become slow
Systems become unstable
From Mall Analogy to Real Kubernetes Components
Just like a Mall Needs
Multiple shops
Backup shops
A management team
A help desk
From Mall Analogy to Real Kubernetes Components
Applications also need:
Transition from Analogy to Technical Concept(Slide 5)
What is Clustering in Kubernetes?
Clustering means multiple machines (Nodes) working together as one single system
Instead of running application on one single server,
We distribute workload across multiple Nodes
Cluster = Group of Nodes managed by Kubernetes
In Kubernetes:
A Cluster consists of:
-
Control Plane
-
Multiple Worker Nodes
How clustering works in Kubernetes:
Nodes join a Cluster
Multiple computer(Nodes) join a cluster
1
Control Plane manages all Nodes
Control Plane manages and oversees all Nodes
2
Scheduler Places Pods
Scheduler decides On which Node Pod should run
3
Pods Rescheduled on failure
If one Node fails:
Pods are rescheduled on another Node
4
Traffic is Distributed
Traffic is evenly balanced across Pods on all Nodes
5
How clustering works in Kubernetes:
Example:
Pod A
Cluster ensures application remains available
Pod A
Pod B
Pod B
Pod C
If Node 2 fails
-
Pod B is recreated on Node 1 or Node 3
What is a Node?
A Node is a worker machine in Kubernetes
It can be :
Physical Machine
Virtual Machine
Cloud Instance
It is where containers actually run
What is a Node?
Kubelet
- Communicates with Control Plane
- Ensures containers are running correctly
Container Runtime
- Runs and manage containers
- Example: containerd , CRI-O
Kube-proxy
- Manage networking inside the node
- Handles traffic routing between Pods
How Node works?
Control Panel
Sends Instructions to Node
1
Node
Kubelet
Receive Instructions
2
Container Runtime
Pulls image and runs container
3
Kube-proxy
Manages networking
4
How Node works?
Node = Execution environment
If Node fails:
Pods rescheduled to another Node
What is a Pod?
Pod is the smallest deployable unit in Kubernetes wrapping around one or more containers
Containers inside a Pod:
Share the same IP address
Share the same network namespace
Can share storage using Volumes
What is a Pod?
Important understanding:
Pod is NOT equal to container.
Pod is a logical application unit that Kubernetes manages.
If your application has:
-
Main container (app)
-
Sidecar container (logging / monitoring)
Both run inside the same Pod
How Pod works:
Scheduler
Assigns Pod to a Node
Pods
Gets a unique IP inside the cluster
Kubelet
Starts the containers
If Container Crashes
It restarted inside same Pod
Important Concept:
Pods are Temporary
What is a ReplicaSet?
ReplicaSet is a controller that ensures:
A fixed number of identical Pod replicas are always running
What is a ReplicaSet?
Example
It always maintains the correct number
- If you define replicas: 3
ReplicaSet ensures:
-
Exactly 3 Pods exist at all times
If Pod Crashes
It creates a new one
If Extra Pod Exist
It Removes them
Why is ReplicaSet Important?
High availability
Fault tolerance
Self-healing
Differentiate Between Labels and Selectors
Attach Identifying metadata to pods
Use to describe Pods
They are Key-value Pair
Filter Pods by matching label
Use to Find Pods by label
They are used for Filtering Conditions
Labels
Selectors
How Does ReplicaSet Work?
Actual State
Running Pods
Desired State
3 Pods
Replicaset
Fixes to match Desire state
Adds Pods
Remove Crash Pods
What is Deployment?
Deployment is a higher-level controller built on top of ReplicaSet
Rolling updates
Rollbacks
Scaling
ReplicaSet
Application updates
Deployment manages:
Application lifecycle manager When you create a Deployment:
When you update container image:
Ensures zero downtime
Deployment
Creates New ReplicaSet
Gradually replaces old Pods
Keeps history of versions
Application lifecycle manager When you create a Deployment:
So,
Text
ReplicaSet
Maintains count
deployment
Manages versions + lifecycle
What is Service?
Problem
Pods have dynamic IP addresses it means it can change
Service provides:
Stable network identity for Pods
What is Service?
Solution: Service gives:
Permanent IP (ClusterIP)
DNS name
Load balancing
How Service works:
1
Service is created with a label selector
2
It finds all Pods matching those labels
3
It creates a stable virtual IP (ClusterIP)
4
Traffic sent to Service IP is forwarded to one of the Pods
Service uses:
kube-proxy
iptables or IPVS rules to distribute traffic
If Pod crashes → where does traffic shift?
Step 2
Kubernetes
removes the failed Pod from Service endpoints
Step 3
Service automatically updates its internal list
Step 4
Traffic is shifted to remaining healthy Pods
Step 1
ReplicaSet creates a new Pod
How Kubernetes Manages Applications
Cluster
Group of Nodes working as one
Node
Machine inside cluster
Pod
Temporary application unit
How Kubernetes Manages Applications
ReplicaSet
Ensures required number of Pods
Deployment
Manages lifecycle and updates
Service
Stable network layer that routes traffic to healthy Pods
Summary
5
Service provides stable access to Pods
4
Deployment manages updates and scaling
3
ReplicaSet maintains the required number of Pods
2
Pod is the smallest deployable unit
1
Node runs the application workload
Quiz
Which component is responsible for rolling updates?
A. ReplicaSet
B. Pod
C. Deployment
D. Node
Quiz
Which component is responsible for rolling updates?
A. ReplicaSet
B. Pod
C. Deployment
D. Node
