Cloud Computing

Undergraduate Study Guide

Covers: Definitions • Service & Deployment Models • Virtualization • Storage • Security • Economics

1. What Is Cloud Computing?

Cloud computing is the on-demand delivery of computing resources — servers, storage, databases,
networking, software, analytics, and intelligence — over the internet ("the cloud") with pay-as-you-go
pricing. Instead of buying and maintaining physical hardware, organisations rent capacity from a cloud
provider.

NIST defines cloud computing as a model for enabling ubiquitous, convenient, on-
Key
demand network access to a shared pool of configurable computing resources that
Definition
can be rapidly provisioned and released with minimal management effort.

Five essential characteristics (NIST)

• users can provision resources automatically without human interaction from the
[Link]-demand self-service —
• services are available over the network and accessed through standard client
[Link] network access —
• provider resources are pooled to serve multiple consumers (multi-tenancy), dynamically
assigned based on [Link] pooling —
• capabilities can be elastically provisioned and released to scale with [Link]
elasticity —
• resource usage is monitored, controlled, and reported — enabling the pay-per-use
[Link] service —

Cloud Computing — Undergraduate Study Guide | Page 1

2. Service Models: IaaS, PaaS, SaaS
Cloud services are typically categorised into three layers, often illustrated as a stack. Each layer
abstracts more complexity from the user.

Model What you manage Examples

IaaS Infrastructure as a Virtualised compute, storage, and AWS EC2, Azure VMs, Google
Service networking. You manage OS Compute Engine, DigitalOcean
upwards.
PaaS Platform as a Managed platform for building & Heroku, Google App Engine,
Service deploying apps. Provider manages Azure App Service, AWS Elastic
OS, runtime, middleware. Beanstalk
SaaS Software as a Fully managed application delivered Gmail, Microsoft 365, Salesforce,
Service over browser. Provider manages Dropbox, Zoom
everything.
FaaS Function as a Run individual functions triggered by AWS Lambda, Azure Functions,
Service events. No servers to manage at all Google Cloud Functions
(serverless).

Think of pizza: IaaS = you buy ingredients and cook. PaaS = you order a base and
Analogy add toppings. SaaS = you order a complete pizza. FaaS = you just eat one slice
when you want it.

3. Deployment Models
Model Description
Public cloud Infrastructure owned and operated by a third-party provider (AWS,
Azure, GCP). Shared across many customers. Lowest cost, highest
scalability.
Private cloud Infrastructure dedicated to a single organisation. Can be on-premise
or hosted. Greater control and security; higher cost.
Hybrid cloud Combination of public and private cloud, interconnected. Workloads
move between them based on need. Common for regulated
industries.
Multi-cloud Using services from two or more cloud providers simultaneously.
Avoids vendor lock-in; increases resilience.
Community cloud Shared infrastructure for a specific community with common
concerns (e.g., government agencies, healthcare providers).

4. Virtualisation

Cloud Computing — Undergraduate Study Guide | Page 2

Virtualisation is the foundation of cloud computing. It creates abstracted, software-defined versions of
physical resources, allowing many virtual machines (VMs) to run on a single physical host.

Hypervisors
A hypervisor (Virtual Machine Monitor) manages VMs on a host. There are two types:
• runs directly on hardware. Examples: VMware ESXi, Microsoft Hyper-V, KVM. Used in
cloud [Link] 1 (bare-metal) —
• runs on top of a host OS. Examples: VirtualBox, VMware Workstation. Used for
development/[Link] 2 (hosted) —

Containers vs. Virtual Machines

Property Virtual Machines Containers
Isolation Full OS per VM — strong isolation Process-level isolation via
namespaces — lighter
Boot time Minutes Seconds or milliseconds
Size Gigabytes (full OS image) Megabytes (app + dependencies
only)
Portability Tied to hypervisor Runs anywhere with container
runtime
Use case Running different OSes, strong Microservices, CI/CD, rapid
security boundaries scaling
Examples VMware, KVM, Hyper-V Docker, containerd, Podman +
Kubernetes

Containers share the host OS kernel — they are NOT full virtual machines. This
Key
makes them fast to start and cheap to run, but they provide weaker isolation than
Insight
VMs. In practice, containers often run INSIDE VMs for the best of both worlds.

5. Cloud Storage Types

Type Description & Use Case
Block storage Raw volumes attached to VMs like a hard drive. High performance,
low latency. Used for databases and OS disks. (e.g., AWS EBS,
Azure Disk)
Object storage Stores unstructured data as objects (files + metadata) accessed via
HTTP/API. Massively scalable, cheap. (e.g., AWS S3, Azure Blob,
GCS)
File storage Traditional file system interface (NFS/SMB) shared across multiple
instances. (e.g., AWS EFS, Azure Files)

Cloud Computing — Undergraduate Study Guide | Page 3

In-memory / cache Extremely fast key-value stores in RAM. Used for sessions, caches,
real-time leaderboards. (e.g., Redis, Memcached, ElastiCache)
Archival storage Lowest cost, highest latency. For backups and compliance data
rarely accessed. (e.g., AWS Glacier, Azure Archive)

6. Cloud Networking Concepts

• logically isolated section of the cloud where you launch resources in a virtual network
you [Link] (Virtual Private Cloud) —
• subdivisions of a VPC; public subnets have internet access, private subnets do
[Link] —
• virtual firewalls controlling inbound/outbound traffic at the instance [Link] Groups
—
• distribute incoming traffic across multiple instances for availability and [Link]
Balancers —
• caches content at edge locations worldwide to reduce latency. (e.g., AWS CloudFront,
Cloudflare)CDN (Content Delivery Network) —
• translates human-readable domain names to IP addresses. Cloud providers offer
managed DNS (Route 53, Azure DNS).DNS —
• private, encrypted connections between on-premise networks and the [Link] / Direct
Connect —

7. Scalability & High Availability

Scaling strategies
Strategy How it works Trade-offs
Vertical scaling (scale Add more power to an existing Simple but has a ceiling; causes
up) machine (more CPU, RAM). downtime; expensive at high end.
Horizontal scaling (scale Add more instances to distribute Preferred in cloud; no downtime;
out) load. near-unlimited scale.
Auto-scaling Automatically add/remove instances Cost-efficient; responds to real-
based on metrics (CPU, time demand changes.
requests/sec).

High availability concepts

• geographic areas with multiple data centres (e.g., us-east-1, europe-west-2).Regions —
• physically separate data centres within a region. Deploy across 2+ AZs to survive a
facility [Link] Zones (AZs) —
• system continues operating even when components [Link] tolerance —

Cloud Computing — Undergraduate Study Guide | Page 4

 • Recovery Time Objective (how long to restore) and Recovery Point Objective (how much
data loss is acceptable).RTO / RPO —

8. Cloud Security
Shared The cloud provider secures the infrastructure (hardware, network, physical
Responsi datacentres). The customer secures what runs on it (OS patches, data encryption,
bility access control, application code).
Model

Key security concepts

• controls who can access which cloud resources. Principle of least privilege: grant only
the permissions [Link] (Identity & Access Management) —
• data stored in cloud storage is encrypted on disk (AES-256 standard).Encryption at rest —
• data moving between services is encrypted using TLS/[Link] in transit —
• requires two or more verification factors to access [Link] (Multi-Factor
Authentication) —
• cloud providers hold certifications like ISO 27001, SOC 2, PCI-DSS, HIPAA-eligible
[Link] & certifications —
• cloud providers offer built-in or managed DDoS mitigation (AWS Shield, Azure DDoS
Protection).DDoS protection —

9. Cloud Economics
Pricing models
Model Description
On-demand Pay by the hour/second for what you use. No commitment. Highest
price per unit.
Reserved instances Commit to 1 or 3 years in exchange for up to 75% discount. Best for
stable workloads.
Spot / Preemptible Bid for spare capacity at up to 90% discount — can be terminated
anytime. Good for batch jobs.
Savings Plans Flexible commitment to a spend level ($/hr) across instance types.
Simpler than reserved.
Free tier AWS, GCP, and Azure offer free-tier usage limits for learning and
prototyping.

Cloud Computing — Undergraduate Study Guide | Page 5

CapEx vs. OpEx
• upfront investment in physical hardware. Fixed cost, long depreciation cycle. Traditional
data [Link] (Capital Expenditure) —
• ongoing operational cost billed as you go. Cloud computing converts IT spend from
CapEx to OpEx, improving cash flow and [Link] (Operational Expenditure) —

Total When comparing cloud vs. on-premise, TCO must include: hardware, power,
Cost of cooling, physical space, IT staff, maintenance, and the opportunity cost of time spent
Ownershi managing infrastructure.
p (TCO)

10. Major Cloud Providers — Service Comparison

Service Category AWS Azure
Compute (VMs) EC2 Azure Virtual Machines
Serverless Lambda Azure Functions
Containers (managed) ECS / EKS AKS (Azure Kubernetes Service)
Object storage S3 Azure Blob Storage
Managed database RDS Azure SQL Database
(SQL)
NoSQL database DynamoDB Cosmos DB
AI / ML platform SageMaker Azure Machine Learning
CDN CloudFront Azure CDN
DNS Route 53 Azure DNS
IAM AWS IAM Azure Active Directory

11. Key Terms Quick Reference

Term Definition
Elasticity Ability to scale resources up or down automatically to match
demand.
Multi-tenancy Multiple customers sharing the same physical infrastructure, isolated
logically.
Hypervisor Software layer that creates and manages virtual machines on
physical hardware.
Namespace Linux kernel feature that isolates processes — the foundation of
containers.

Cloud Computing — Undergraduate Study Guide | Page 6

cgroups Linux control groups that limit and monitor resource usage per
process group.
Microservices Architectural style where an app is built as small, independent,
deployable services.
Serverless Execution model where the provider manages all infrastructure; you
deploy functions only.
SLA Service Level Agreement — the uptime/performance guarantee from
the provider (e.g., 99.99%).
Latency Time for a request to travel from user to server and back.
Throughput Amount of data processed per unit of time.
Idempotency Operation that produces the same result no matter how many times
it is executed.
Infrastructure as Code (IaC) Managing cloud infrastructure through code/config files (Terraform,
CloudFormation) rather than manual UI.

This guide covers core concepts aligned with standard undergraduate cloud computing curricula. For deeper study, refer to
AWS Skill Builder, Microsoft Learn, or Google Cloud Training.

Cloud Computing — Undergraduate Study Guide | Page 7