Introduction
Cloud computing service models define how much control you have over your infrastructure versus how much is managed by the cloud provider. Understanding these models is crucial not just for passing certification exams, but for making informed architectural decisions in real-world projects.
Infrastructure as a Service (IaaS)
IaaS provides the most control. You're essentially renting virtual hardware.
What You Manage:
- Operating system patches and updates
- Runtime environments
- Application code
- Data
- Network configuration (within your VNet/VPC)
What the Provider Manages:
- Physical hardware
- Hypervisor
- Physical networking
- Data center operations
Real-World Example: Running a legacy application that requires specific OS configurations. You need full control over the environment, so you spin up VMs on Azure or AWS.
When to Choose IaaS:
- Lift-and-shift migrations
- Custom software requiring specific OS versions
- Applications with regulatory requirements for infrastructure control
- Development/test environments
Popular IaaS Services:
Azure: Virtual Machines • Blob Storage • Managed Disks • VNet • Load Balancer
AWS: EC2 • S3 • EBS • VPC • ELB
GCP: Compute Engine • Cloud Storage • VPC
Platform as a Service (PaaS)
PaaS abstracts the infrastructure, letting you focus on code.
What You Manage:
- Application code
- Data
- Application configuration
What the Provider Manages:
- Everything else (OS, runtime, scaling, patching)
Real-World Example: Deploying a web application to Azure App Service or AWS Elastic Beanstalk. You push your code, and the platform handles everything else.
When to Choose PaaS:
- New application development
- Microservices architectures
- When you want to minimize operational overhead
- Rapid prototyping and MVP development
Popular PaaS Services:
Azure: App Service • Azure SQL • Cosmos DB • Azure Functions
AWS: Elastic Beanstalk • RDS • DynamoDB • Lambda
GCP: App Engine • Cloud SQL • Firestore • Cloud Functions
Software as a Service (SaaS)
SaaS is fully managed software you consume.
What You Manage:
- Configuration and settings
- User access and permissions
- Your data (within the application)
What the Provider Manages:
- Everything about the application
Real-World Example: Microsoft 365, Salesforce, or Google Workspace. You use the software; the provider handles all infrastructure and application maintenance.
Popular SaaS Examples:
- Productivity: Microsoft 365, Google Workspace
- CRM: Salesforce, HubSpot
- Communication: Slack, Zoom, Teams
- DevOps: GitHub, GitLab, Jira
The Shared Responsibility Model
Understanding service models is directly tied to the shared responsibility model:
Responsibility Spectrum
IaaS ←——————————————————————————→ SaaS
More Customer Control More Provider Control
┌──────────────┬─────────────┬─────────────┬─────────────┐
│ │ IaaS │ PaaS │ SaaS │
├──────────────┼─────────────┼─────────────┼─────────────┤
│ Data │ Customer │ Customer │ Customer │
│ Applications │ Customer │ Customer │ Provider │
│ Runtime │ Customer │ Provider │ Provider │
│ OS │ Customer │ Provider │ Provider │
│ Virtualization│ Provider │ Provider │ Provider │
│ Hardware │ Provider │ Provider │ Provider │
└──────────────┴─────────────┴─────────────┴─────────────┘
Decision Framework
Use this flowchart to choose the right model:
- Do you need full OS control? → IaaS
- Are you building custom applications? → PaaS
- Do you just need to use software? → SaaS
- Legacy app with specific dependencies? → IaaS
- New cloud-native development? → PaaS or Serverless
Cost Analysis Example
Scenario: Hosting a web application with database
- IaaS (VMs): $200/mo infrastructure + 20 ops hours @ $50 = $1,200 total
- PaaS: $350/mo infrastructure + 5 ops hours @ $50 = $600 total
- SaaS: $500/mo subscription + 1 ops hour @ $50 = $550 total
Note: PaaS often wins when you factor in operational costs!
Beyond the Exam: Real Architecture Decisions
In practice, most cloud architectures use a mix of all three models:
- IaaS for legacy systems or specialized workloads
- PaaS for application code and databases
- SaaS for productivity tools and specialized services
Cost Considerations:
- IaaS: Pay for uptime (even when idle), but potentially lower per-hour costs
- PaaS: Often more expensive per unit, but you pay less for operations staff
- SaaS: Subscription-based, predictable, but less flexibility
Exam Tips
Common exam questions test:
- Identifying which service model fits a scenario
- Understanding who is responsible for what
- Recognizing service examples (e.g., "Azure SQL Database is PaaS")
- Shared responsibility model boundaries
Watch for keywords:
- "Full control over OS" → IaaS
- "Focus on code, not infrastructure" → PaaS
- "Ready-to-use application" → SaaS
Key Takeaway
The exam will test your knowledge of these models. Real-world success comes from knowing when each model is appropriate and how they work together in a hybrid architecture.