Cloud Concepts exam cheat sheet¶
Scenario → answer¶
| Scenario or clue | Best match |
|---|---|
| Use cloud resources without buying physical servers | Public cloud |
| Environment dedicated to one organization | Private cloud |
| Azure connected to an on-premises datacenter | Hybrid cloud |
| Azure plus another public cloud provider | Multicloud |
| Buy servers upfront | CapEx |
| Pay for ongoing consumption | OpEx |
| Increase CPU or RAM on one VM | Scale up / vertical scaling |
| Add more VM instances | Scale out / horizontal scaling |
| Adjust capacity dynamically with demand | Elasticity |
| Maximize uptime | High availability |
| Recover and continue after failure | Reliability |
| Control and patch the operating system | IaaS |
| Deploy an app without managing the OS | PaaS |
| Use a ready-made application | SaaS |
| Run code in response to an event | Serverless / Azure Functions |
| Separated datacenters within one region | Availability zones |
| Associated Azure regions | Region pair |
| Directory of identities | Microsoft Entra tenant |
| Apply governance to several subscriptions | Management group |
| Billing, quota, and access boundary | Subscription |
| Group resources with one lifecycle | Resource group |
| Provider measures usage by service meters | Measured service / consumption model |
| Minimize interruption during ordinary failures | High availability |
| Recover in another region after major disruption | Disaster recovery |
| Recover historical data after deletion or corruption | Backup |
Three fast comparisons¶
Scale¶
- Up/down changes the size of one resource.
- Out/in changes the number of resources.
Service model¶
- IaaS: configure the server.
- PaaS: publish the application.
- SaaS: use the software.
IaaS → customer patches the guest OS
PaaS → provider patches OS and managed runtime
SaaS → provider operates the application
All three → customer controls data, identities, and access decisions
Location¶
- Datacenter: physical building.
- Availability zone: isolated datacenter group inside a region.
- Region: geographic deployment area.
- Region pair: relationship used by some services; not automatic replication of everything.
Azure hierarchy¶
Tenant = identities
Management group = govern subscriptions together
Subscription = billing, quotas, access scope
Resource group = lifecycle container
Resource = actual Azure service instance
Availability¶
- One instance: can remain a single point of failure.
- Multiple instances: can protect against instance failure when traffic and state are designed correctly.
- Availability zones: protect against zone-level failure inside a region.
- Multiple regions: support regional disaster recovery.
- Backup: preserves recoverable historical state; redundancy alone is not backup.
Dangerous absolute words¶
Question statements containing always, all, automatically, or guarantees deserve extra attention. Common false claims include:
- Cloud always costs less.
- SaaS transfers all security responsibility to Microsoft.
- A region pair automatically replicates every resource.
- One VM placed in a zone is automatically zone resilient.
- A resource group's region forces all its resources into that region.
- A private cloud is simply any virtualized server.
- A subscription is an availability boundary.
- Tags automatically flow to all child resources.