AZ-900: Core Cloud Concepts
The foundational cloud ideas the AZ-900 exam tests — service models, deployment models, and the economics of cloud.
Cloud computing means renting computing resources — servers, storage, databases — over the internet, paying only for what you use, instead of buying and running your own hardware.
The three service models describe how much the provider manages versus you:
- IaaS (Infrastructure as a Service): you rent virtual machines and networks. You manage the OS and apps; Azure manages the physical hardware. Most control.
- PaaS (Platform as a Service): you deploy code; Azure manages the OS, runtime, and patching. Balanced.
- SaaS (Software as a Service): you just use finished software, like Microsoft 365. Least management.
The three deployment models describe where things run:
- Public cloud — shared Azure infrastructure.
- Private cloud — dedicated to one organization.
- Hybrid cloud — a mix, connecting on-premises with the public cloud.
graph LR subgraph IaaS A1["You: OS, runtime, app"] A2["Azure: hardware, network"] end subgraph PaaS B1["You: app + data"] B2["Azure: OS, runtime, scaling"] end subgraph SaaS C1["You: just use it"] C2["Azure: everything"] end
The exam emphasizes the business benefits:
- CapEx → OpEx: swap big up-front hardware purchases (capital expenditure) for pay-as-you-go operating costs.
- Elasticity & scalability: add capacity in minutes for a traffic spike, then scale back down.
- High availability & reliability: built-in redundancy across datacenters.
- Global reach: deploy close to users worldwide.
- Shared responsibility: Azure secures the cloud; you secure what you put in it.
A startup launching a web app chooses cloud because:
- They can’t predict traffic, so elasticity lets them scale on demand.
- They have no money for servers, so OpEx billing means near-zero upfront cost.
- They want users in Europe and Asia served quickly, so global regions help.
They pick PaaS (Azure App Service) so they deploy code without managing VMs — trading some control for far less operational work.
- For each of these, decide IaaS, PaaS, or SaaS: an Azure Virtual Machine; Azure SQL Database; Outlook on the web.
- Explain in one sentence why pay-as-you-go is “OpEx” rather than “CapEx”.
- Give a real scenario where a hybrid deployment makes sense.
- List two benefits of the cloud that matter most to a small team and why.
Cheat Sheet▾
| Concept | Quick definition |
|---|---|
| IaaS | Rent VMs/networks; you manage OS + app |
| PaaS | Deploy code; Azure manages the platform |
| SaaS | Use finished software; manage nothing |
| Public cloud | Shared provider infrastructure |
| Private cloud | Dedicated to one org |
| Hybrid cloud | Mix of on-prem + public |
| CapEx | Up-front capital spend on hardware |
| OpEx | Ongoing pay-as-you-go operating cost |
| Elasticity | Scale up/down with demand |
| High availability | Stays up via redundancy |
| Shared responsibility | Security split between you & Azure |
Common Interview Questions▾
What's the difference between IaaS, PaaS, and SaaS?
They differ by how much you manage. IaaS gives you VMs (you manage OS + app); PaaS gives you a managed platform (you manage just the app); SaaS gives you ready software (you manage nothing).
What does 'CapEx to OpEx' mean in cloud economics?
Instead of large up-front capital purchases of hardware (CapEx), cloud shifts you to ongoing, usage-based operating expenses (OpEx) — you pay only for what you use.
Explain the shared responsibility model.
Security duties are split: the provider secures the underlying cloud (physical hosts, network), while the customer secures what they run on it (data, identities, access, app configuration). The exact split depends on the service model.
Troubleshooting Scenarios▾
Mixing up “scalability” and “elasticity”. Scalability is the ability to grow; elasticity is doing it automatically up and down in response to demand. The exam rewards the precise term.
Confusing high availability with disaster recovery. HA keeps a system running through small failures (redundancy). DR is recovering after a major outage (backups, failover regions). They solve different problems.
Assuming “cloud” always means cheaper. Cloud reduces upfront cost and adds flexibility, but poorly managed resources can cost more. Right-sizing and turning off unused resources matters.