What to know before you read.
- Begin with architecture concepts that transfer across providers.
- Security, identity, observability and cost belong in the first project—not after it.
- Learn one provider deeply enough to deploy a complete system before comparing others.
Cloud is an operating model, not a list of products
Cloud computing gives teams on-demand access to computing, storage, networking and managed services through programmable platforms. The durable skill is choosing and operating those resources responsibly.
A service catalogue can make the field look larger than it is. Organise it around recurring questions: where will the workload run, where will data live, who can access it, how will failures be detected, how will it scale and what will it cost?
Stage 1: infrastructure, networks and identity
Understand regions, availability zones, virtual machines, containers, serverless execution, object and block storage, databases, virtual networks, subnets, routing and load balancing at a conceptual level.
Study identity and access alongside them. Learn least privilege, service identities, secrets and the shared-responsibility model. A publicly exposed resource is not a harmless beginner mistake when real data is involved.
- Compute: where code executes
- Storage: how data is persisted
- Networking: how components communicate
- Identity: who or what may act
- Reliability: how the system responds to failure
- Cost: what usage is billable
Stage 2: choose one provider and translate the concepts
Pick AWS, Microsoft Azure or Google Cloud using your employer, project or learning environment as the signal. Map each foundation concept to one provider service and note where managed services reduce operational work.
Set budgets, alerts and tagging conventions before deployment. Cost awareness is part of architecture, not an administrative task for later.
| Week | Focus | Deliverable |
|---|---|---|
| 1 | Compute, storage and regions | Provider-neutral architecture sketch |
| 2 | Networks and identity | Private/public boundary and access map |
| 3 | Deploy a small workload | Reachable application with controlled access |
| 4 | Add managed data | Backup and data-flow explanation |
| 5 | Logs, metrics and alerts | Observable failure scenario |
| 6–8 | Reliability, cost and capstone | Documented end-to-end deployment |
Stage 3: operate what you deploy
A successful first deployment is the beginning of cloud learning. Add logs, health checks, alerts, backups and a simple recovery procedure. Then inspect the bill and identify the resources that create cost.
Test one failure deliberately. Stop an instance, break a permission in a safe environment or simulate an unavailable dependency. Observe what the user sees and what the operator learns.
Stage 4: document one complete architecture
Your capstone should include a diagram, deployment instructions, identity model, data path, observability plan, estimated cost and improvement backlog. Keep credentials and secrets out of the repository.
This evidence demonstrates architecture thinking regardless of provider. After completion, map the components to another cloud to understand what transfers and what changes.
- Use infrastructure naming and tags consistently.
- Separate configuration from secrets.
- Record assumptions and capacity limits.
- Explain why managed or self-managed components were chosen.
- Clean up resources after the project.
Choose foundations or a provider specialisation
Begin with Cloud Computing Fundamentals if the architecture vocabulary is new. Choose AWS, Microsoft Azure or Google Cloud when you are ready to work inside one provider ecosystem.
The goal of 30 focused hours is not mastery of every service. It is a coherent foundation and one project you can operate and explain.
Questions readers often ask.
Which cloud concept should I learn first?+
Begin with compute, storage, networking and identity, then connect them in a small architecture.
Do I need Linux for cloud computing?+
Basic command-line and operating-system knowledge helps, but it can be learned alongside your first deployments.
How do I avoid surprise cloud charges?+
Use budgets and alerts, understand billable resources, tag experiments and delete everything you no longer need.
Decision guide