Service Models

IaaS vs PaaS vs SaaS: Cloud Service Models Compared With Examples

IaaS vs PaaS vs SaaS compared: what each model manages, who uses it, real examples from AWS, Azure, Google Cloud and Salesforce, and how to choose.

On this page
  1. Short answer
  2. What the abbreviations stand for
  3. The three models at a glance
  4. Who manages what
  5. IaaS: control and flexibility
  6. PaaS: focus on the application
  7. SaaS: finished software
  8. Serverless and other models
  9. Examples of each model
  10. Comparing control, effort and cost
  11. How to choose
  12. Common misconceptions
  13. Market size and forecasts
  14. Frequently asked questions

Short answer

IaaS, PaaS and SaaS are the three main cloud service models. They differ in how much of the stack the provider manages. With IaaS, you rent infrastructure and manage everything from the operating system up. With PaaS, you deploy your application on a managed platform. With SaaS, you use a finished application. The more the provider manages, the less control you have and the less work you do.

What the abbreviations stand for

  • IaaS: Infrastructure as a Service
  • PaaS: Platform as a Service
  • SaaS: Software as a Service
  • FaaS: Function as a Service, the model behind many serverless products

The most widely used reference for the three models is NIST Special Publication 800-145, published in September 2011.

The three models at a glance

IaaS PaaS SaaS
What you get Virtual machines, storage and networks A platform to build, deploy and run applications A finished application over the internet
What you manage Operating system, middleware, runtime, data and applications Your application and data Your data, user accounts, access and settings
What the provider manages Networking, storage, servers and virtualization Everything up to the runtime The whole stack
Typical users Infrastructure and operations teams Developers End users and business teams
Examples Amazon EC2, Azure Virtual Machines, Google Compute Engine, DigitalOcean Droplets Heroku, Google App Engine, AWS Elastic Beanstalk, Azure App Service Salesforce, Microsoft 365, Google Workspace, Slack, Zoom, Dropbox

Who manages what

A cloud stack has several layers: networking, storage, servers, virtualization, the operating system, middleware, runtime, data and applications. Each model moves the line between your responsibility and the provider’s.

  • In IaaS, the provider runs the layers from networking through virtualization. You manage the operating system and everything above it.
  • In PaaS, the provider runs everything through the runtime. You manage your application and your data.
  • In SaaS, the provider runs the whole stack. You still manage your data, user accounts, access settings and configuration.

Security is shared in all three models. The provider secures the underlying platform, but you remain responsible for what you control, such as data, identities and access settings.

IaaS: control and flexibility

IaaS gives you the building blocks of cloud computing: virtual servers, storage, networks and load balancers. You choose the operating system, install your own software and configure the network. This model suits workloads that need a custom environment, existing applications moved to the cloud with few changes, and teams that want full control.

Billing is usually per instance hour or second, plus storage and data transfer out of the cloud. You are responsible for sizing, patching and monitoring.

PaaS: focus on the application

PaaS gives developers a managed environment for running code. The provider handles the operating system, the runtime and much of the scaling, so you deploy code and data and the platform runs it. Many PaaS products can add or remove instances as demand changes.

PaaS is not always cheaper than IaaS. At steady, high load, the markup on a managed platform can cost more than running your own virtual machines. For how PaaS developed, read the evolution of PaaS. To check whether a particular service is PaaS or IaaS, read IaaS or PaaS? Azure Storage, App Engine and Elastic Beanstalk.

Two specialized managed services sit close to PaaS. Backend as a service (BaaS) is usually classed as a subset of PaaS. Database as a service (DBaaS) is usually counted as PaaS too, though some vendors argue it is not. Read what backend as a service is and what database as a service is.

SaaS: finished software

SaaS delivers a complete application through a browser or app. The provider runs the infrastructure, updates the software and maintains the application. Subscriptions are usually priced per user per month. You manage the users, the access settings and the configuration, along with the data you put into the application.

SaaS is the easiest model to start with, but it gives you the least control over the software itself. For more examples and case studies, see SaaS examples and case studies. For a comparison with installed software, see SaaS vs on-premise software.

Serverless and other models

Serverless computing, or Function as a Service (FaaS), lets you run code in response to events without managing servers. AWS Lambda, launched in 2014, is the best-known example. Azure Functions and Google Cloud Run functions work in a similar way. You pay per request and for the duration of each run. Lambda’s maximum timeout is 15 minutes.

Serverless still runs on servers, so “no servers” is misleading. The provider manages them, and you do not.

Containers as a service (CaaS), such as Amazon EKS, Azure Kubernetes Service (AKS) and Google Kubernetes Engine (GKE), sit between IaaS and PaaS. The boundaries between categories are not sharp. Managed Kubernetes, Elastic Beanstalk and object storage are classified differently by different sources.

Examples of each model

Product Model Note
Amazon EC2 IaaS Virtual servers you configure
Azure Virtual Machines IaaS Virtual machines on Azure
Google Compute Engine IaaS Virtual machines on Google Cloud
Heroku PaaS Deploy applications without managing servers
Google App Engine PaaS Standard and Flexible environments
AWS Elastic Beanstalk PaaS Deploys your code to managed AWS resources
Salesforce SaaS Customer relationship management
Microsoft 365 SaaS Email, documents and collaboration
Gmail (part of Google Workspace) SaaS Email application
Slack SaaS Team messaging

Some providers offer more than one model. Microsoft offers SaaS (Microsoft 365), PaaS (Azure App Service) and IaaS (Azure Virtual Machines). AWS offers services in all three categories.

Comparing control, effort and cost

Factor IaaS PaaS SaaS
Control Highest Medium Lowest
Operational effort Highest Medium Lowest
Time to start Longest Medium Shortest
Typical billing Per instance, plus storage and data transfer Per instance or per usage Usually per user per month
Portability Highest, because you control the stack Lower; platform services differ by provider Depends on the vendor’s data export options

How to choose

  • Choose IaaS if you need control of the operating system, custom networking, or software that does not fit a managed platform.
  • Choose PaaS if you are building your own application and want to spend less time on infrastructure.
  • Choose SaaS if you need a finished tool for a business task.

Many organizations use all three. A company might run email as SaaS, host its web application on PaaS and keep a legacy system on IaaS.

Common misconceptions

  • “Cloud means SaaS.” It does not. SaaS is one of three service models within cloud computing. See what SaaS vs cloud means.
  • “PaaS is always cheaper.” Not always. Steady high load can make a managed platform more expensive than IaaS.
  • “Serverless means no servers.” The provider runs the servers, so you do not manage them.
  • “The provider handles security.” Responsibility is shared. You still manage data, identities and access.

Market size and forecasts

Analysts publish forecasts for each category, and the numbers are revised often. Gartner’s forecast for 2025 put public cloud spending at about $723 billion, with SaaS the largest segment at about $299 billion, IaaS at about $212 billion and PaaS at about $209 billion. Check the latest forecast, and note its date, before you quote a figure.

Frequently asked questions

What are the three main categories of cloud computing services?

IaaS, PaaS and SaaS. NIST Special Publication 800-145 defines these three service models, along with five essential characteristics and four deployment models.

What are the five pillars of cloud computing?

“Pillars” is not a formal term, and sources use it loosely. The five essential characteristics that NIST lists are on-demand self-service, broad network access, resource pooling, rapid elasticity and measured service.

Is AWS a SaaS or IaaS?

AWS offers all three models. Core services such as Amazon EC2 are IaaS, many managed services are PaaS, and AWS also offers SaaS-style products. It is best thought of as a provider across all three models rather than as one.

Is Netflix a SaaS or PaaS?

Netflix is closest to SaaS, as a consumer streaming service. It is not a PaaS, because it does not offer a platform that developers can use to build their own applications. Classifications vary between sources. See whether Netflix is a SaaS or PaaS for details.

Is Gmail a PaaS or SaaS?

Gmail is SaaS. It is a finished application that people use through a browser or app, and Google runs the whole stack.

Keep exploring

Guides from other topics

All topics →