Cloud computing is defined by three service models that describe how much of the stack the provider manages. IaaS gives you raw infrastructure, PaaS adds a managed platform, and SaaS delivers a finished product.
The choice is a trade-off between control and convenience. Each model moves responsibility up or down the stack.
Infrastructure as a Service (IaaS)
Virtual machines, storage and networks that you provision and manage yourself. You control the operating system, security and scaling.
- Most control, most operational responsibility.
- Good for legacy migrations, custom infrastructure, and workloads with strict compliance needs.
- Examples: AWS EC2, Azure Virtual Machines, GCP Compute Engine.
Platform as a Service (PaaS)
A managed runtime that runs your code without you managing servers. Scaling, patching and load balancing are handled for you.
- Best for application developers who want to move fast.
- Examples: AWS Elastic Beanstalk, Google Cloud Run, Heroku.
Software as a Service (SaaS)
A finished application the provider runs and maintains. You consume the product, not the infrastructure.
- Examples: Gmail, Slack, Notion.
Most organizations use all three: SaaS for commodity tools, PaaS for their own applications, and IaaS for the workloads that need raw control.
Which should you choose?
Start with SaaS for off-the-shelf needs. Use PaaS for building apps. Reserve IaaS for when a requirement genuinely forces it. Every step up the stack saves you time and operational risk.
Cloud models FAQ
What is Function as a Service?
FaaS is a subset of PaaS where you deploy single functions that run on demand. AWS Lambda and Cloud Functions are the classic examples.
Is PaaS more expensive than IaaS?
The list price is higher, but total cost is usually lower because you avoid the engineering time spent managing infrastructure.
