Chapter 1 – Cloud vs On‑Prem & Fundamentals

What is Cloud Computing?

• Definition: A model that lets you access a shared pool of computing resources (servers, storage,
apps, services) on‑demand, over the network, with minimal management effort.
• Think of it like: Electricity. You don’t build a power plant; you flip a switch and pay for what you
use.

What is On‑Premises (On‑Prem)?

• Definition: Your organization buys, installs, and runs all hardware/software in its own data
center (or server room). You manage power, cooling, networking, security, upgrades, scaling,
and staffing.
• Analogy: Owning your own car (buy, maintain, insure) vs. using a ride‑hailing app (cloud).

Cloud vs On‑Prem:
Aspect Cloud Computing On‑Premises
Cost model OpEx (pay‑as‑you‑go/subscription) CapEx + OpEx (hardware purchase +
ongoing)

Provisioning Minutes (APIs/console) Weeks–months (procure, rack, configure)

speed

Scalability Elastic/auto scale Limited by hardware; scale requires buying

more

Responsibility Shared (provider manages a lot) You manage everything end‑to‑end

Upgrades Continuous by provider You plan and execute

Typical fit Variable or fast‑growing needs, Steady workloads, strict data‑residency,

global apps legacy/latency needs
Pros & Cons of Cloud & On-Premises Services
Category Pros Cons
Cloud - Fast to start; global reach; elastic - Ongoing OpEx; egress/licensing
scaling - Managed services reduce surprises without FinOps - Vendor lock‑in
undifferentiated heavy lifting - risk; service limits/quotas - Data
Built‑in resiliency across Availability residency/compliance may require
Zones careful design
On‑Prem - Full control/custom hardware and - Slow to scale; large CapEx; refresh every
data locality - Predictable 3–5 years - You own HA/DR, patching,
performance inside your LAN - security stack - Requires data center
Potentially lower cost for very space, power, cooling
steady, high‑utilization workloads

Deployment Models
• Public Cloud: Shared provider infrastructure for general use (e.g., AWS, Azure, GCP).
• Private Cloud: Cloud tech dedicated to one org (on‑prem or hosted).
• Community Cloud: Shared by several organizations with common concerns (e.g., compliance,
mission).
• Hybrid Cloud: Combine two or more clouds (e.g., on‑prem + public) with portability.
• Multi‑Cloud (strategy): Using services from multiple public clouds (e.g., AWS + Azure + GCP) to
reduce risk, leverage best‑of‑breed services, or negotiate costs.

Service Models
• IaaS (Infrastructure as a Service): Provider manages hardware, networking, virtualization. User
manages: Operating system, middleware, runtime, data, and applications.
Example: AWS EC2, Azure VMs, Google Compute Engine.

• PaaS (Platform as a Service): Provider manages hardware, networking, virtualization, OS,

middleware, and runtime. User manages: Data and applications only.
Example: AWS Elastic Beanstalk, Azure App Service, Google App Engine.

• SaaS (Software as a Service): Provider manages everything (infrastructure, OS, apps, data). User
manages: Just usage and basic configurations.
Example: Salesforce, Google Workspace, Microsoft 365.

• On-Prem (for contrast): Organization manages everything: facilities, power, networking,

hardware, OS, middleware, runtime, data, and applications.
Characteristics Of Cloud Computing
• On‑demand self‑service: Users can automatically provision resources (like VMs or storage)
whenever they need, without human help from the provider.
• Broad network access: Services are available over the internet and can be accessed from
anywhere, using different devices (laptops, mobiles, tablets).
• Resource pooling: The provider pools computing resources to serve multiple customers, with
isolation and security, similar to different tenants sharing an apartment building.
• Rapid elasticity: Resources can quickly scale up or down according to demand — almost
instantly, giving the sense of infinite capacity.
• Measured service: Usage is monitored, metered, and billed based on consumption (like
electricity or water).

AWS Global Infrastructure

• Regions: Large geographic areas (like Mumbai, Ohio, Sydney). Each Region has multiple data
centers.
• Availability Zones (AZs): Independent data centers inside a Region with separate power, cooling,
and networking. Regions usually have at least 3 AZs for high availability.
• Edge Locations: Smaller sites spread worldwide to deliver content faster (used by CloudFront
CDN and Route 53 DNS).
• Local Zones: Extension of Regions, placed closer to large population or industry centers to
reduce latency.
• Wavelength Zones: Infrastructure embedded inside telecom networks to provide ultra‑low
latency for mobile/5G apps.
• Outposts: AWS infrastructure installed inside a customer’s own data center to run AWS services
on‑prem while still connecting to the AWS cloud.
Current public counts: ~37 Regions / 117 AZs (check AWS Global Infrastructure page for updates).

Cloud Providers: AWS vs Azure vs GCP

Key Differences AWS Azure GCP
Strengths Largest service catalog; Tight Microsoft Strong data/ML (BigQuery);
mature ecosystem; integration; many global network;
strong multi‑AZ defaults regions; licensing developer‑friendly
synergies

Watch‑outs / Pricing complexity; Licensing complexity; Smaller catalog; some

Limitations service sprawl; egress naming overlaps; zone services regional only;
costs availability varies migration re‑architecture

Best fit use‑cases Broad workloads Enterprises already Analytics‑heavy, ML/AI

needing variety; using Microsoft stack; workloads;
enterprise and startup hybrid setups developer‑centric apps
scale
Core AWS Services
• Compute: Services that provide processing power to run applications (EC2, Lambda,
ECS/Fargate, EKS).

• Storage: Services to store and access data reliably (S3, EBS, EFS, FSx).

• Databases: Managed database services for structured and unstructured data (RDS, DynamoDB,
Aurora, Redshift).

• Networking: Services that connect and secure your applications (VPC, Route 53, ELB/ALB/NLB,
CloudFront).

• Security & Identity: Services to manage access and protect data (IAM, KMS, Secrets Manager,
Cognito).

• Observability: Tools to monitor, trace, and audit activity (CloudWatch, X-Ray, CloudTrail).

• Integration: Services to connect applications and workflows (SQS, SNS, EventBridge, Step
Functions).

• Analytics & AI: Tools for analyzing data and adding machine learning (Athena, Glue, EMR,
QuickSight, Bedrock, SageMaker).

Examples
• Cloud vs On‑Prem analogy: Cloud is like renting an apartment; on‑prem is building your own
house.
• When cloud wins: New app with spiky traffic → auto scale, pay for surge.
• When on‑prem makes sense: Legacy ERP with steady load and high egress cost.

Interview Questions
1. What are the 5 essential characteristics of cloud computing?

o On‑demand self‑service, broad network access, resource pooling, rapid elasticity,

measured service.
2. Differentiate IaaS vs PaaS vs SaaS with examples.

o IaaS: you manage OS/apps (EC2). PaaS: you deploy code; provider runs runtime/OS
(Elastic Beanstalk). SaaS: full app delivered (Salesforce).
3. Public vs Private vs Hybrid vs Community cloud?

o Public: provider‑owned for general use; Private: dedicated to one org; Community: shared
by orgs with common needs; Hybrid: mix with portability.
4. What is an AWS Region vs Availability Zone?

o Region = geographic area; AZ = one or more discrete data centers in a Region.

5. Why can cloud be cheaper—or not?
 o Cheaper due to elasticity/managed services; not if governance is weak (zombie resources,
egress). Use pricing/TCO tools.
6. Give real‑world fit examples.

o Startups with unpredictable load → cloud.

o Regulated workloads with steady utilization → private/hybrid.