CLF-C02 · AZ-900 · AIF-C01
IaaS vs PaaS vs SaaS explained with examples
The difference between infrastructure, platform and software as a service, who manages what in each, and how AWS and Azure exams test it.
Three ways to rent computing
IaaS, PaaS and SaaS describe how much of the stack the cloud provider runs for you. The more the provider runs, the less you control and the less you have to maintain.
A useful picture is a meal. IaaS is renting a kitchen: you buy ingredients and cook. PaaS is a meal kit: the ingredients arrive prepared and you assemble them. SaaS is ordering at a restaurant: you just eat.
Infrastructure as a service (IaaS)
With IaaS you rent virtual machines, storage and networks. The provider runs the physical datacenter, servers and virtualization layer. You install and patch the operating system, runtime, middleware and application, and you configure the firewall rules.
Examples are Amazon EC2 and Azure Virtual Machines. IaaS fits lift-and-shift migrations, where a company moves existing servers to the cloud with few changes, and test environments that need a specific OS build or custom software.
IaaS gives the most control and the most work. If a question says the company needs full control of the operating system, or needs to install a legacy agent on the server, IaaS is the answer.
Platform as a service (PaaS)
With PaaS the provider also runs the operating system and runtime. You bring your code or your data and configure the service. Patching, scaling hooks and most of the underlying maintenance belong to the provider.
Examples include AWS Elastic Beanstalk, Amazon RDS, Azure App Service and Azure SQL Database. A developer pushes a web app to App Service and never logs in to a server. A database team creates an RDS instance and AWS handles engine patching and automated backups.
PaaS fits teams that want to build applications without managing servers. The trade-off is less control: you usually cannot pick the exact OS version or install arbitrary software on the host.
Software as a service (SaaS)
SaaS is a finished application delivered over the internet. You sign in, configure settings and use it. The provider runs everything underneath, including the application code.
Microsoft 365 email and Office apps are the standard example on AZ-900. Other common SaaS products are web-based CRM, payroll and file-sharing services. You still own your data, your user accounts and who gets access, even though you never see a server.
Who manages what
This comparison shows how responsibility moves from you to the provider as you go from IaaS to SaaS.
- Always the provider: physical datacenters, physical network, physical hosts.
- Always you: your data, your devices, your user accounts and identities.
- IaaS: you manage the OS, network controls, runtime and application.
- PaaS: the provider manages the OS and runtime; you manage the application, its settings and access.
- SaaS: the provider manages the application too; you manage configuration, data and users.
Where serverless fits
Serverless services such as AWS Lambda and Azure Functions are usually grouped with PaaS. You upload a function, it scales automatically, and you pay only while it runs. There are still servers, you just never see or manage them.
On the AWS AI Practitioner exam, the same idea shows up with managed AI services. Calling a foundation model through Amazon Bedrock is closer to SaaS or PaaS than hosting the model yourself on EC2 GPU instances, which is IaaS.
Exam tips
Match keywords to the model. "Full control", "custom OS" and "lift and shift" point to IaaS. "Focus on code", "no server management" and "managed database" point to PaaS. "Ready-to-use application", "email" and "no development" point to SaaS.
Also expect questions that combine this topic with the shared responsibility model: the customer's share is largest in IaaS and smallest in SaaS, but it never drops to zero because data and identities stay with the customer.
Practice questions
A company wants a serverless service that can discover the schema of its datasets, store that metadata in a central catalog, and run extract, transform, and load (ETL) jobs. Which AWS service should it use?
- AWS Quick Sight
- Amazon Polly
- Amazon Kinesis Data Streams
- AWS Glue
Answer
D. AWS Glue is a serverless data integration service with crawlers, the AWS Glue Data Catalog, and managed ETL jobs. AWS Quick Sight visualizes data in dashboards but does not catalog or transform it.
CLF-C02 question 259
A company wants a single pricing commitment that also lowers the cost of its serverless workloads. Which services are covered by Compute Savings Plans? (Choose two.)
- AWS Fargate
- AWS Lambda
- Amazon S3
- Amazon Route 53
- Amazon DynamoDB
Answer
A, B. Compute Savings Plans apply to Amazon EC2, AWS Fargate and AWS Lambda usage, regardless of instance family or Region. Storage, NoSQL database and DNS services are not covered by Compute Savings Plans.
CLF-C02 question 300
A company wants to lift and shift hundreds of physical and virtual servers to Amazon EC2 using continuous block-level replication and minimal cutover downtime. Which AWS service should it use?
- AWS Transfer Family
- AWS Database Migration Service (AWS DMS)
- AWS Snowball Edge
- AWS Application Migration Service
Answer
D. AWS Application Migration Service replicates source servers continuously into AWS and launches them as EC2 instances, which makes it the main tool for rehosting. AWS DMS migrates databases, not entire servers.
CLF-C02 question 211
What does the phrase "security in the cloud" refer to in the AWS shared responsibility model?
- AWS's duty to secure its global network
- AWS's duty to secure its data centers
- The customer's duty to secure what it puts in AWS
- AWS's duty to patch managed service hosts
Answer
C. Security in the cloud is the customer's side: its data, IAM settings, network configuration and, for IaaS, the operating system. The other choices describe security of the cloud, which is AWS's side.
CLF-C02 question 105
Practice CLF-C02 Practice AZ-900 Practice AIF-C01 All topics