4.1 Cloud computing
Cloud computing delivers computing services over the internet or through a cloud environment. Instead of buying and operating every server yourself, you use resources as needed.
Cloud models
| Model |
Meaning |
Best fit |
Common trap |
| Public cloud |
Provider-owned cloud resources shared across customers with logical isolation |
Fast deployment, broad cloud services, consumption-based usage |
“Public” does not mean your data is automatically public |
| Private cloud |
Cloud environment dedicated to one organization |
Dedicated control or specialized internal requirements |
Private cloud is not the same as a single private endpoint |
| Hybrid cloud |
Combines on-premises or private-cloud resources with public-cloud services |
Gradual migration, legacy dependencies, mixed environments |
If any required workload remains on-premises while Azure is also used, consider hybrid |
Consumption-based model
You use resources when needed and generally pay according to usage. The key advantage is avoiding unnecessary upfront hardware purchases.
| Term |
Meaning |
Exam clue |
| Capital expenditure (CapEx) |
Upfront purchase of assets such as physical servers |
“Buy hardware before deployment” |
| Operational expenditure (OpEx) |
Ongoing spending such as a variable cloud bill |
“Monthly bill changes with usage” |
| Consumption-based pricing |
Charges align with resource usage |
“Stop paying after deleting or stopping resources” |
Shared responsibility model
Cloud adoption changes who manages each layer.
| Layer |
On-premises |
IaaS |
PaaS |
SaaS |
| Physical datacenter |
Customer |
Microsoft |
Microsoft |
Microsoft |
| Physical network and hosts |
Customer |
Microsoft |
Microsoft |
Microsoft |
| Guest OS |
Customer |
Customer |
Microsoft |
Microsoft |
| Runtime and middleware |
Customer |
Customer |
Microsoft |
Microsoft |
| Application |
Customer |
Customer |
Customer |
Mostly Microsoft |
| Data, identities, and access decisions |
Customer |
Customer |
Customer |
Customer |
Rule: the more managed the service, the less infrastructure the customer manages. The customer still remains responsible for data governance, identity choices, and access decisions.
Serverless
Serverless computing lets you run code without managing servers. The cloud provider handles infrastructure management and scaling details.
Exam clue: event-driven code, short execution, trigger-based processing, no server management → Azure Functions.