Cloud Computing,
from first principles to AWS & Azure.

What is Cloud Computing?

The one-line definition: Cloud computing is renting computing resources (servers, storage, databases, networking, software) over the internet, paying only for what you use, instead of buying and maintaining your own hardware.

The Official NIST Definition

The US National Institute of Standards and Technology (NIST) defines cloud computing as a model for enabling on-demand network access to a shared pool of configurable computing resources (networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort.

The Indian Context — Why It Matters Now

• IRCTC handles 10+ lakh ticket bookings daily. During Tatkal, the load spikes 50× in 10 minutes. Without cloud's elasticity, the servers would crash.
• UPI processed over 16 billion transactions in a single month in 2024. The NPCI backend uses cloud infrastructure to scale instantly.
• JioCinema streamed IPL 2024 finals to 6+ crore concurrent viewers — only possible because cloud auto-scales.
• Zomato & Swiggy see 5× traffic during dinner hours. They use cloud auto-scaling so customers don't get "service unavailable".

Cloud Paradigms

A "paradigm" simply means a way of thinking. Cloud computing changes the way we think about IT in three big ways:

The 5 Essential Characteristics of Cloud

NIST identifies 5 essential characteristics that something must have to be called "cloud". If even one is missing, it's not real cloud computing.

The Benefits — Why Companies Move to Cloud

The Big 3 Cloud Vendors

Globally, three companies dominate the cloud market. Together they hold about 65% of global cloud spend.

Service Name Mapping — A Quick Reference

Why On-Premise Infrastructure Struggles

"On-premise" (or "on-prem") means servers physically located in your own office or data centre. Before cloud, this was the only option. Here's why it struggles in the modern world:

The Capacity Planning Trap

Imagine you run a food delivery app like Swiggy. Your traffic looks like this:

• 3 AM to 11 AM — low traffic (10% of peak)
• 1 PM & 8 PM — peak traffic (100%)
• Festive day (Diwali) — 3× peak (300%)

If you buy on-premise servers, you must size them for the worst case — 300%. That means for 95% of the year, 70% of your hardware sits idle, eating electricity and depreciating. You paid ₹2 crore for hardware you barely use.

The 6 Hard Problems of On-Premise

SaaS — Software as a Service

Definition: SaaS is software delivered over the internet on a subscription basis. You don't install it, you don't update it, you don't host it. You just log in and use it.

Traditional Software vs SaaS

Famous SaaS Examples (Indian + Global)

IaaS — Infrastructure as a Service

Definition: IaaS gives you the raw building blocks of computing — virtual machines, storage, networks, firewalls — over the internet. You install the OS, you install the database, you write the application. The provider just gives you the empty hardware.

What You Get with IaaS

• Virtual Machines (VMs) — like a remote computer you rent (AWS EC2, Azure VM)
• Storage — block storage (like a hard drive) and object storage (S3, Blob)
• Networking — virtual networks, firewalls, load balancers
• Operating System — you choose: Ubuntu, Amazon Linux, Windows Server, RHEL

Famous IaaS Examples

PaaS — Platform as a Service

Definition: PaaS gives you a ready-made platform to run your applications. You write code; the platform handles OS, runtime, scaling, patching, load balancing. You don't manage servers at all.

Famous PaaS Examples

Virtualization — The Engine of Cloud

Definition: Virtualization is the technology that allows one physical server to be split into many virtual servers, each running its own OS, completely isolated from the others. This is the magic that makes cloud possible.

Why Cloud Needs Virtualization

• Multi-tenancy — Many customers can safely share one big server
• Efficient hardware use — A 64-core server can run 30 VMs instead of sitting 90% idle
• Isolation — If one customer's VM crashes, others are unaffected
• Fast provisioning — Creating a VM takes 60 seconds (creating a physical server takes weeks)
• Mobility — VMs can be moved between physical machines without users noticing

Types of Virtualization

Hypervisor — The Boss of VMs

Definition: A hypervisor (also called a Virtual Machine Monitor or VMM) is the software layer that creates, runs, and manages virtual machines. It sits between hardware and VMs, deciding which VM gets which CPU cycles and RAM.

VM Provisioning & Migration Services

What is VM Provisioning?

Provisioning means setting up and configuring a virtual machine so it's ready to use. The steps are:

1. Select template — Choose an OS image (Ubuntu 22.04, Windows Server 2022, Amazon Linux)
2. Configure resources — Pick CPU (vCPUs), RAM (GB), disk size (GB), network
3. Apply security — Set firewall rules, SSH keys, IAM roles
4. Boot the VM — Hypervisor allocates resources and starts the OS
5. Install software — Either manually via SSH/RDP, or automatically via scripts (cloud-init, Ansible)

In AWS, this entire process takes 60-120 seconds and is fully self-service. You can do it via Console (GUI), CLI, SDK, or Infrastructure-as-Code (Terraform, CloudFormation).

VM Migration Services

Migration means moving a running VM from one physical host to another with little or no downtime. This is essential for cloud providers — they constantly rebalance VMs across hardware.

Cloud Deployment Models

There are 4 ways to deploy cloud — depending on who owns it and who can access it:

Private Cloud Deployment — A Closer Look

Building a private cloud means setting up cloud-like services inside your own data centre. You provide self-service VMs, storage, networking — but only for your own organization.

Popular Private Cloud Software:

• OpenStack — Open-source, used by ISRO, Walmart, Comcast
• VMware vSphere/vCloud — Enterprise standard, used by most Indian banks
• Microsoft Azure Stack — Azure-compatible private cloud
• Red Hat OpenShift — Container-focused private cloud

Challenges of Cloud Environment

Cloud isn't all rainbows. Here are the real challenges every cloud engineer faces:

Administering & Monitoring Cloud Services

"If you can't measure it, you can't manage it." Cloud monitoring tells you what your infrastructure is doing — CPU usage, response times, errors, costs.

What to Monitor

Monitoring Tools

Cloud Monitoring vs Traditional Monitoring

Deploying an Application Over Cloud

Let's walk through deploying a typical Flask application (something you build in your PG-DBDA labs) to AWS:

Architecture for a Production Flask App

Step-by-Step Deployment Flow

1. Develop locally — Write Flask app on your laptop
2. Push to GitHub — Version control your code
3. Provision infrastructure — Create EC2 instance(s), RDS database, S3 bucket via Console or Terraform
4. Set up VPC + security groups — Network and firewall rules
5. SSH into EC2 — Install Python, dependencies (or use Docker)
6. Clone repo & run app — Use Gunicorn + Nginx as production server
7. Add load balancer & auto-scaling — Handle multiple users, scale automatically
8. Add CloudFront CDN — Speed up static asset delivery
9. Set up monitoring & logging — CloudWatch alarms, log groups
10. Set up CI/CD — GitHub Actions auto-deploys on every git push

IaaS vs PaaS vs SaaS — Side by Side

The famous "pizza analogy" makes this concept stick. Imagine you want pizza:

The Pizza Analogy

Cloud Pricing Models

How do you actually get billed? The five main models:

Sample Pricing (AWS Mumbai region, approximate)

Tip: Always set up Billing Alerts in AWS so you get an email if monthly spend crosses ₹500.

The Shared Responsibility Model

This is one of the most important concepts in cloud — and the most misunderstood. Security in the cloud is shared between you and the provider.

The Simple Rule

• AWS/Azure/GCP is responsible for security OF the cloud — physical buildings, hardware, hypervisor, network cables.
• You are responsible for security IN the cloud — your data, your IAM users, your firewall rules, your OS patches, your application code.

The shared responsibility shifts depending on the model:

* Even in SaaS, your account data, sharing settings, and user access are still your responsibility.

Introduction to AWS

Amazon Web Services launched in 2006. Today it's the world's largest cloud provider, with 200+ services across 33 regions globally.

Core AWS Concepts

AWS Service Categories

EC2 — Elastic Compute Cloud

What it is: EC2 lets you rent virtual servers (called instances) in the cloud. Think of it as your remote Linux/Windows computer in AWS data centre.

EC2 Instance Types

AWS offers hundreds of instance types, categorized by what they're optimized for:

Naming Convention Decoded

An instance like t3.micro breaks down as:

• t = family (T-series, burstable general purpose)
• 3 = generation (3rd gen)
• micro = size (nano < micro < small < medium < large < xlarge < 2xlarge…)

So m6i.2xlarge = General-purpose, 6th gen, Intel, 2× extra-large.

EC2 Storage Options

• EBS (Elastic Block Store) — Like a hard drive attached to your VM. Persists when VM is stopped. Types: gp3 (general), io2 (high-IOPS), st1 (throughput).
• Instance Store — Physically attached disk. Very fast but data is lost when VM stops. Used for caching/temp data.
• EFS — Shared file system (NFS) that multiple EC2s can mount.

Lab: Create & Access an EC2 Instance

Step-by-Step: Launch Linux EC2

1. Sign in to AWS Console → search "EC2"
2. Click "Launch Instance"
3. Name: my-first-server
4. AMI: select Amazon Linux 2023 (Free Tier eligible)
5. Instance type: t2.micro or t3.micro (Free Tier)
6. Key pair: Create new → name it my-key → download .pem file (keep it safe!)
7. Network settings: Allow SSH (port 22) from your IP only, HTTP (80) from anywhere
8. Click "Launch Instance"

Connect via SSH (from Linux/Mac)

# Set correct permissions on key file
chmod 400 my-key.pem

# SSH into the instance
ssh -i my-key.pem ec2-user@<your-public-ip>

# Once inside, update packages
sudo dnf update -y

# Install Apache web server
sudo dnf install httpd -y
sudo systemctl start httpd
sudo systemctl enable httpd

# Create a sample page
echo "<h1>Hello from EC2!</h1>" | sudo tee /var/www/html/index.html

# Now open http://<your-public-ip> in browser

Launch Windows Server VM

Same steps but:

• Select AMI: Microsoft Windows Server 2022 Base
• Network: allow RDP (port 3389)
• After launch: click "Connect" → "RDP client" → download .rdp file
• Get password: paste your .pem private key to decrypt the Administrator password
• Open .rdp file → enter password → you're in Windows desktop!

S3 — Simple Storage Service

What it is: S3 is AWS's object storage service. Store anything — files, images, videos, backups, datasets — with virtually unlimited capacity, 99.999999999% (11 nines!) durability.

Key Concepts

• Bucket — A container for objects. Name must be globally unique (e.g., vineeta-pgdbda-bucket-2026). Has a region.
• Object — A file + metadata. Max size: 5 TB. Identified by a key (the filename + path).
• Key — The unique path: uploads/students/2026/photo.jpg
• URL — Every object gets a URL: https://<bucket>.s3.<region>.amazonaws.com/<key>

S3 Storage Classes

Lifecycle Policies: Automatically move objects between classes. E.g., "Move logs to Glacier after 90 days, delete after 2 years."

Lab: Create an S3 Bucket

Create a Bucket via Console

1. AWS Console → search "S3" → "Create bucket"
2. Bucket name: globally unique (e.g., vineeta-cdac-demo-2026)
3. Region: ap-south-1 (Mumbai)
4. Block all public access: keep checked (default — secure)
5. Versioning: enable (allows recovery of deleted/overwritten files)
6. Default encryption: SSE-S3 (enabled by default now)
7. Create bucket

Upload & Manage Files via AWS CLI

# Install AWS CLI and configure
aws configure
# Enter Access Key, Secret Key, region: ap-south-1

# Upload a single file
aws s3 cp ./report.pdf s3://vineeta-cdac-demo-2026/reports/

# Upload entire folder
aws s3 cp ./project/ s3://vineeta-cdac-demo-2026/project/ --recursive

# List buckets
aws s3 ls

# List objects in a bucket
aws s3 ls s3://vineeta-cdac-demo-2026/ --recursive

# Download a file
aws s3 cp s3://vineeta-cdac-demo-2026/reports/report.pdf .

# Sync a folder (incremental, like rsync)
aws s3 sync ./local-folder s3://vineeta-cdac-demo-2026/backup/

# Delete an object
aws s3 rm s3://vineeta-cdac-demo-2026/old-file.txt

Host a Static Website on S3

1. Create index.html with your portfolio content
2. Create bucket (e.g., vineeta-portfolio.com)
3. Uncheck "Block all public access" (this one should be public)
4. Properties → Static Website Hosting → Enable → Index document: index.html
5. Permissions → Bucket Policy → allow public s3:GetObject
6. Upload index.html
7. Visit the website URL — your site is live!

AWS Lambda — Serverless Computing

What it is: Lambda lets you run code without provisioning servers. You upload a function, AWS runs it when triggered (HTTP request, S3 upload, schedule, etc.), and you pay only for actual execution time.

How Lambda Works

1. You write a function (Python, Node.js, Java, Go, .NET, Ruby)
2. Upload it to Lambda
3. Attach a "trigger" — what causes it to run? (HTTP, S3 upload, DynamoDB change, cron schedule, SQS message…)
4. When triggered, AWS spins up a container, runs your code, returns the result, shuts down
5. You pay for compute time used (per millisecond)

Lambda Pricing

• Requests: ~₹16 per 1 million requests
• Compute: ~₹1.4 per million GB-seconds
• Free tier: 1 million requests + 400,000 GB-seconds per month, always free

This means small projects can run completely free. A Telegram bot with 10,000 messages/month? ₹0.

Common Lambda Use Cases

• Image processing — When user uploads to S3, Lambda creates thumbnails
• API backend — API Gateway + Lambda = fully serverless REST API
• Scheduled tasks — Daily backup of database to S3 (replaces cron jobs)
• IoT data processing — Stream sensor data through Lambda
• Chatbots — WhatsApp / Telegram bots
• ETL — Triggered when new file lands in S3, process it

Lab: Create a Lambda Function

Hello World Lambda (Python)

1. AWS Console → search "Lambda" → "Create function"
2. Author from scratch · Name: hello-pgdbda · Runtime: Python 3.12 · Architecture: x86_64
3. Click "Create function"
4. In the code editor, paste:

import json
from datetime import datetime

def lambda_handler(event, context):
    name = event.get('name', 'PG-DBDA Student')
    now = datetime.now().strftime('%H:%M:%S')
    
    return {
        'statusCode': 200,
        'body': json.dumps({
            'message': f'Hello {name}!',
            'timestamp': now,
            'from': 'AWS Lambda (Mumbai)'
        })
    }

5. Click "Deploy"
6. Click "Test" → create test event with {"name": "Vineeta"} → Test
7. See the JSON response in the output panel!

Add an HTTP Trigger (Function URL)

1. Configuration tab → Function URL → Create
2. Auth type: NONE (for testing)
3. You get a public URL like: https://abc123.lambda-url.ap-south-1.on.aws/
4. Open it in browser → your Lambda runs over HTTP!
5. Pass query string: ?name=Vineeta (you'll need to modify the handler to read query strings)

VPC — Virtual Private Cloud

What it is: A VPC is your own private, isolated network inside AWS. Think of it as your own data centre — but virtual. You define the IP range, subnets, routing, and firewall rules.

VPC Components

Typical VPC Architecture (3-Tier)

Lab: Create a VPC

Easy Path — VPC Wizard

1. AWS Console → VPC → "Create VPC"
2. Select "VPC and more" (creates everything in one click)
3. Name: pgdbda-vpc
4. IPv4 CIDR: 10.0.0.0/16
5. Number of AZs: 2
6. Public subnets: 2, Private subnets: 2
7. NAT gateways: 1 (or "None" to save cost during lab)
8. VPC endpoints: S3 Gateway (free!)
9. Create VPC

In 2 minutes, AWS creates VPC, subnets, IGW, route tables, NACLs — everything wired up correctly.

Introduction to Microsoft Azure

Azure is Microsoft's cloud platform, launched in 2010. It's the #2 cloud provider globally, dominant in enterprises that already use Windows Server, Active Directory, SQL Server, or Office 365.

Why Azure Wins in Enterprises

• Hybrid Cloud — Azure Arc, Azure Stack lets you extend on-prem to cloud seamlessly
• Windows Integration — Native Active Directory, Group Policy, .NET
• Office 365 Sync — One Microsoft login for everything
• Enterprise Agreements — Easier procurement for big Indian banks/PSUs

Azure Core Concepts

Three Ways to Manage Azure

Key Azure Services

Azure Data Services

Azure Storage

Azure Functions

Azure's equivalent of AWS Lambda. Serverless code execution triggered by events (HTTP, Blob upload, timer, queue message).

Languages: C#, JavaScript, Python, Java, PowerShell, TypeScript, Go.

Pricing: Consumption plan — first 1 million executions free per month.

# Sample Python Azure Function (HTTP trigger)
import azure.functions as func

def main(req: func.HttpRequest) -> func.HttpResponse:
    name = req.params.get('name', 'World')
    return func.HttpResponse(
        f"Hello, {name}!",
        status_code=200
    )

Lab: Deploy from GitHub

Deployment to AWS via GitHub Actions

The modern way: push code to GitHub, a workflow auto-deploys to AWS.

# .github/workflows/deploy.yml
name: Deploy to AWS

on:
  push:
    branches: [main]

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@v4
        with:
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ap-south-1

      - name: Deploy to S3
        run: aws s3 sync ./public s3://my-bucket/ --delete

      - name: Invalidate CloudFront cache
        run: aws cloudfront create-invalidation \
          --distribution-id ${{ secrets.CF_DIST_ID }} \
          --paths "/*"

Deployment to Azure via GitHub Actions

name: Deploy to Azure Web App

on:
  push:
    branches: [main]

jobs:
  build-and-deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: '3.11'

      - name: Install dependencies
        run: pip install -r requirements.txt

      - name: Deploy to Azure Web App
        uses: azure/webapps-deploy@v3
        with:
          app-name: my-flask-app
          publish-profile: ${{ secrets.AZURE_PUBLISH_PROFILE }}

Deployment via Azure DevOps

Azure DevOps Pipelines is an alternative to GitHub Actions, popular in enterprises:

1. Create Azure DevOps account at dev.azure.com
2. Create a new Project & import your code from GitHub
3. Pipelines → Create pipeline → Select repo → Choose template (Python web app, Node.js, etc.)
4. The YAML pipeline auto-generates
5. Configure service connection to your Azure subscription
6. Save & run — pipeline builds, tests, and deploys

Quick Revision Sheet

One-Line Definitions (memorize these)

The 5 NIST Characteristics (memorize)

1. On-demand self-service
2. Broad network access
3. Resource pooling
4. Rapid elasticity
5. Measured service

The 4 Deployment Models

1. Public
2. Private
3. Hybrid
4. Community

The 3 Service Models

1. IaaS (Infrastructure)
2. PaaS (Platform)
3. SaaS (Software)

Common Interview Questions

Conceptual Questions

1. What is cloud computing? Give 3 real-world examples.
   Cloud computing is on-demand delivery of IT resources over the internet with pay-as-you-go pricing. Examples: Gmail (SaaS), AWS EC2 (IaaS), Heroku (PaaS).
2. Difference between scalability and elasticity?
   Scalability is the ability to handle growth (manual or planned). Elasticity is automatic, real-time scaling up/down based on current demand.
3. Why would a bank choose private cloud over public?
   Compliance (RBI data localization), full control over data, predictable performance, regulatory requirements.
4. Explain the shared responsibility model.
   Cloud provider is responsible for security of the cloud (hardware, hypervisor). Customer is responsible for security in the cloud (data, IAM, configurations).
5. What's the difference between vertical and horizontal scaling?
   Vertical = bigger server (more CPU/RAM). Horizontal = more servers (more instances behind a load balancer). Horizontal is preferred in cloud.
6. What is a hypervisor? Difference between Type 1 and Type 2?
   A hypervisor manages VMs. Type 1 runs directly on hardware (ESXi, used in cloud). Type 2 runs on a host OS (VirtualBox, used on laptops).
7. Why is virtualization important for cloud computing?
   It enables multi-tenancy, efficient hardware use, fast provisioning, isolation between customers — all the foundations of cloud economics.

AWS-Specific Questions

1. Difference between EC2 and Lambda?
   EC2 = persistent virtual server you manage. Lambda = stateless function that runs only when triggered, fully managed by AWS.
2. What is an S3 bucket?
   A container for objects (files) in S3. Globally unique name, lives in one region, contains unlimited objects.
3. Difference between EBS and S3?
   EBS = block storage attached to one EC2 (like a hard drive). S3 = object storage accessible via HTTP from anywhere.
4. What is a VPC and why use it?
   A private network within AWS. Lets you isolate resources, control IP addressing, set firewall rules, and segment public/private workloads.
5. Public vs private subnet?
   Public subnet has a route to internet via Internet Gateway. Private subnet doesn't — for databases & backend services.
6. What is Auto Scaling?
   A service that automatically adds/removes EC2 instances based on demand or schedule.

Scenario Questions

1. Your e-commerce site crashes during Big Billion Days sale. How would you fix it on AWS?
   Put EC2 behind an Auto Scaling Group + Application Load Balancer. Add CloudFront CDN for static assets. Use RDS Multi-AZ. Set CloudWatch alarms.
2. How would you reduce a high AWS bill?
   Identify idle resources via Cost Explorer. Buy Reserved Instances for steady workloads. Use Spot for batch jobs. Move infrequent S3 data to Glacier. Delete unused EBS volumes & snapshots. Set budget alerts.
3. You need to process 1 TB of CSV files nightly. Which AWS services?
   Store data in S3. Use AWS Glue or EMR (Spark) for processing. Trigger via EventBridge schedule. Output to S3/Redshift. For smaller files, Lambda can work too.