Cloud fundamentals: AWS, Azure and Google Cloud
Learn what the cloud really is, the handful of building blocks every provider offers, and how AWS, Azure and Google Cloud name them. You will be able to read any provider's service list without feeling lost.
"The cloud" means renting computing resources (servers, storage, databases, networks) from a provider over the internet and paying for what you use, instead of buying and running your own machines. Think of electricity: you do not build a power station, you plug in and get a bill.
Three providers dominate: Amazon Web Services (AWS), Microsoft Azure and Google Cloud (GCP). Others exist (including Indian providers), but the ideas below are the same everywhere.
Why companies use it
Own servers versus cloud
| Question | Own servers | Cloud |
|---|---|---|
| Starting cost | Buy hardware first | Pay as you go, start small |
| Getting a new server | Weeks (order, install) | Minutes (a few clicks or one command) |
| Growing for a sale or exam-result day | Buy for the peak, idle the rest | Add capacity when needed, remove later |
| Who fixes hardware | You | The provider |
| Main risk | Under or over buying | A surprise bill, or a misconfigured service |
The cloud is not automatically cheaper or safer. It moves work: you no longer replace a failed disk, but you must now design for cost, permissions and failure yourself.
The shared responsibility model
The provider secures the cloud itself: buildings, hardware, the virtualisation layer. You secure what you put in it: who has access, whether a storage bucket is public, whether your servers are patched, whether your data is encrypted. Most real cloud security incidents come from customer configuration mistakes, such as a public bucket or a leaked access key, not from the provider being hacked.
Three service models
IaaS, PaaS and SaaS
| Model | You manage | Provider manages | Example |
|---|---|---|---|
| IaaS (Infrastructure) | OS, runtime, app, data | Hardware, network, virtualisation | A virtual server (EC2) |
| PaaS (Platform) | Your app and data | OS, runtime, scaling | App Service, Elastic Beanstalk, Cloud Run |
| Serverless (functions) | Only your code | Servers entirely, scaling to zero | AWS Lambda, Azure Functions |
| SaaS (Software) | Only your usage | Everything | Gmail, Microsoft 365, Salesforce |
Moving down the table you manage less and get less control. For a small API, PaaS or serverless is often the fastest route; for special software or full control, a virtual server.
Geography: regions and availability zones
A region is a geographic area with several data centres, for example Mumbai or Hyderabad in India. An availability zone (AZ) is one or more separate data centres inside a region with their own power and network. Spreading your app across two AZs means one data centre failing does not take you offline.
Choose the region by: distance to your users (lower latency), the law (some data must stay in India, see the DPDP Act and your client's rules), service availability and price, which differs between regions. Azure and Google Cloud use similar ideas, with some differences in naming. Always check each provider's current region list.
The building blocks and their names
Every provider sells the same few categories. Learn the category first; the product name is just a label.
Rough equivalents (check each provider for exact differences)
| Category | AWS | Azure | Google Cloud |
|---|---|---|---|
| Virtual servers | EC2 | Virtual Machines | Compute Engine |
| Object storage (files) | S3 | Blob Storage | Cloud Storage |
| Managed SQL database | RDS | Azure SQL / Database for PostgreSQL | Cloud SQL |
| Serverless functions | Lambda | Functions | Cloud Run functions |
| Run containers | ECS, EKS (Kubernetes) | Container Apps, AKS | Cloud Run, GKE |
| Private network | VPC | Virtual Network | VPC |
| Identity and permissions | IAM | Microsoft Entra ID and Azure RBAC | IAM |
| Monitoring | CloudWatch | Azure Monitor | Cloud Monitoring |
Names change over time (Google renamed some serverless products, for example). When in doubt, open the provider's documentation and search the category.
Core ideas in every cloud
Ideas you will meet in every provider
- Compute: virtual machines, containers or functions that run your code.
- Storage: object storage for files, block disks for servers, managed databases for records.
- Networking: a private network, subnets, firewalls (security groups) and a way for users to reach you.
- Identity: users, roles and policies that decide who can do what. The most important topic for security.
- Elasticity: adding and removing capacity automatically.
- Pay for use: you are billed by time, size or requests, so forgotten resources keep costing money.
A worked example. Brew & Bean wants its menu API online. A simple design: the React site files in object storage behind a CDN, the API in a container service or a function, orders in a managed Postgres database, and images in object storage. Every piece is a category from the table; only the product names differ between providers.
Which provider should you learn?
Start with one, and AWS is a common choice because of its market share and the number of tutorials and jobs. Azure is strong in companies that already use Microsoft products. Google Cloud is popular for data, AI and Kubernetes work. The concepts transfer: someone who understands IAM, VPCs and object storage on AWS can pick up Azure in a few weeks. Look at job posts in your target city to decide.
Cloud accounts need a payment method
All three providers ask for a card even for free plans. Free offers change often, and some services are never free. Set a budget alert on day one and delete what you create when you finish a lab. The lesson *Cloud costs, free tiers and avoiding surprise bills* covers this in detail.
How this comes up in interviews
Common questions: IaaS versus PaaS versus SaaS with examples; what a region and an availability zone are and why you would use two zones; the shared responsibility model; why use object storage instead of a server disk; and how you would host a simple web app on a cloud. Answer with the categories, then name one provider's product for each.
Try it: Map your own project
Take any project you have built (or the Brew & Bean site). List each part (frontend files, API, database, uploaded images, scheduled job). Next to each, write the cloud category it would use and the product name on AWS and on one other provider.
Key takeaways
- The cloud is rented computing, billed by use; it moves work rather than removing it.
- Providers sell the same categories: compute, storage, database, network, identity, monitoring.
- Regions are geographic areas; spread across availability zones to survive a data-centre failure.
- You are responsible for access, configuration and data; the provider secures the hardware.
- Learn one provider properly; the concepts carry over to the others.
Check yourself
Finishing the quiz marks the lesson as done.
1. Which is an example of PaaS?
2. Why deploy across two availability zones?
3. A company's storage bucket was public and data leaked. Under shared responsibility, who is most likely responsible?
4. Which pair are equivalent object-storage services?
5. What is the best habit when you start with a cloud account?
Saved on this device. Create a free account to keep it on every device.
Practise this with Sam
A real mock interview, out loud, with an honest scorecard. Free to start.