Cloud computing concepts make up roughly 25% of the AZ-900 exam. Memorizing terms alone is not enough. You need to understand intuitively what the cloud is, why organizations choose it, and how its cost structure differs from traditional IT. This guide walks through each core concept using everyday analogies.
What Is Cloud Computing?
A hundred years ago, every factory had to own its own generator. They bought the generator, hired technicians, and repaired it when it broke down. If production increased but the generator lacked capacity, the factory had to stop. Electricity was a luxury reserved for the wealthy.
Today? You plug into an outlet and electricity flows. You never need to know where the power plant is or how it operates. You pay only for what you use. In summer, when air conditioning runs more, your electricity bill rises. In winter, when you use less, it falls.
Cloud computing works exactly the same way. Companies like Microsoft, Google, and Amazon operate massive data centers around the world. You access their computing resources (servers, storage, networking, software) over the internet, whenever you need them. Just as you use electricity without buying a generator, you use computing power without buying servers.
Microsoft's official definition: "Delivery of computing services — including servers, storage, databases, networking, software, analytics, and intelligence — over the internet."
Traditional (On-Premises) vs Cloud Comparison
| Item | Traditional (On-Premises) | Cloud | |------|--------------------------|-------| | Server procurement | Buy outright (weeks to months) | Provision in minutes | | Maintenance | Your own IT team | Provider (Microsoft) manages | | Handling demand | Must forecast in advance, risk over/under-investment | Adjust instantly with demand | | Upfront cost | Hundreds of millions in investment | No upfront investment | | Payment model | Purchase then depreciate | Pay only for what you use | | Global expansion | Must build local data centers | Deploy worldwide with a click |
Three Cloud Deployment Models
The cloud comes in three types depending on who owns it and who can use it. Distinguishing between these three is a frequent exam topic.
Public Cloud
Picture an apartment complex. The building (servers, network equipment, data center) is built, owned, and managed by the developer — Microsoft. Each unit (company or individual) uses their own space independently. You live in the same building, but you cannot see what your neighbors are doing. The walls separate everything completely.
The essence of public cloud is that infrastructure is shared among many customers. Microsoft Azure, AWS, and Google Cloud are the major examples.
Advantages: Zero upfront investment Deploy services globally in minutes Pay only for what you use No infrastructure maintenance worries Access to nearly unlimited resources when needed
Disadvantages: No physical control over server location or hardware Some regulatory environments (healthcare data, financial records) may restrict public cloud use
Private Cloud
Think about owning a house. You own the property, control who comes in completely, set every lock and fence as you like. But you bear all costs of purchasing and maintaining the house.
A private cloud is a dedicated cloud environment for a single organization. The organization either owns its own data center or leases an entirely isolated space. Banks, hospitals, military organizations, and government agencies with the strictest security and compliance requirements tend to choose this.
Advantages: Complete control and ownership over infrastructure Highest levels of security and isolation Can meet the most stringent compliance requirements Easier integration with existing legacy systems
Disadvantages: High upfront costs for hardware, installation, and maintenance Requires a dedicated IT team Elasticity is limited compared to public cloud
Hybrid Cloud
Imagine a company running headquarters (private) and branch offices (public) simultaneously. Confidential executive meetings and sensitive documents happen at headquarters; routine sales work and customer service happen at branch offices. The two locations are connected and exchange data as needed.
Hybrid cloud uses public and private cloud together. For example, a hospital keeps patient medical records on a strictly regulated private server, while the appointment booking app and hospital website run on Azure public cloud.
Advantages: Choose the optimal environment based on data sensitivity Preserve existing on-premises investments while gaining cloud benefits Compliance-permitted data goes public; sensitive data stays private
Disadvantages: Increased complexity in connecting and managing two environments Must maintain consistent security policies across private and public
Comparison Summary
| Item | Public Cloud | Private Cloud | Hybrid Cloud | |------|-------------|--------------|-------------| | Owner | Microsoft/Provider | The organization | Mixed | | Cost structure | OpEx (usage-based) | CapEx-heavy | Mixed | | Security control | Provider + customer shared | Full customer control | Varies by zone | | Compliance | Generally suitable | Suited for strict regulations | Flexible combination | | Scalability | Nearly unlimited | Physical limits exist | Elastic on public side | | Analogy | Apartment | Own house | Headquarters + branches |
!3 cloud deployment models
Shared Responsibility Model
This concept appears very frequently on the AZ-900 exam and must be understood precisely.
Back to the apartment building. What is the building manager (Microsoft) responsible for? The elevator, hallway lighting, exterior waterproofing, parking lot security, building electrical systems, and plumbing. Everything related to the building structure itself is the manager's responsibility.
What is the tenant (customer) responsible for? What happens inside the unit, who gets a key to the front door, whether the windows are left open, where valuables are stored. The tenant's behavior and data are the tenant's responsibility.
The cloud shared responsibility model follows exactly this principle. Some things are always Microsoft's responsibility. Some things are always the customer's responsibility. And the line shifts depending on the service model (IaaS, PaaS, SaaS).
Detailed Responsibility Areas
| Responsibility Area | Always Provider (Microsoft) | Always Customer | Depends on Service Model | |--------------------|-----------------------------|-----------------|--------------------------| | Data center physical security (building, locks, guards) | Provider | - | - | | Power supply and cooling systems | Provider | - | - | | Physical network equipment (routers, switches) | Provider | - | - | | Physical server hardware | Provider | - | - | | Data content and classification | - | Customer | - | | Access account and identity management | - | Customer | - | | User devices (laptops, smartphones) | - | Customer | - | | Operating System (OS) patching | - | - | Depends on service type | | Network firewall configuration | - | - | Depends on service type | | Application code | - | - | Depends on service type | | Runtime environment | - | - | Depends on service type |
Key principle: The further you move toward SaaS, the more Microsoft handles. The closer you stay to IaaS, the broader the customer's responsibility. Regardless of which cloud model you use, data content and access account management are always the customer's responsibility.
CapEx vs OpEx — Two Fundamentally Different Cost Structures
Consider buying a car. You pay 30,000 USD upfront. It is a large sum, but the car is yours afterward. You still pay for fuel, insurance, and repairs, but the purchase cost is already settled. That 30,000 USD is CapEx (Capital Expenditure).
Now consider car-sharing (Zipcar, car2go). You pay an hourly rate each time you drive. No need for a 30,000 USD upfront payment. In months with many business trips, costs are higher. In work-from-home months, costs drop. This usage-based spending is OpEx (Operational Expenditure).
Detailed CapEx vs OpEx Comparison
| Item | CapEx (Capital Expenditure) | OpEx (Operational Expenditure) | |------|----------------------------|-------------------------------| | Full name | Capital Expenditure | Operational Expenditure | | Payment | Large upfront lump sum | Pay by usage — monthly/hourly | | IT example | Buy servers, storage, network equipment | Cloud service subscriptions, usage fees | | Cost predictability | Fixed at purchase, then depreciated | Varies with demand, flexible within predictable range | | Flexibility | Hard to change after purchase | Scale up or down at any time | | Accounting | Recorded as asset, depreciated | Expensed immediately | | Cloud relevance | On-premises model | Cloud model |
Consumption-Based Model
The cloud's core philosophy in one sentence: "You pay only for what you use." This is the consumption-based model.
The cloud converts IT spending from CapEx to OpEx. Instead of buying servers, you rent them. Fixed costs become variable costs. When demand rises, costs rise. When demand falls, costs fall.
Benefits of the consumption-based model: Costs are near zero when there is no traffic You can respond flexibly to unexpected demand spikes Investing only what is needed reduces wasted resources The cost of experimenting with new services is low — if it fails, the loss is minimal
In traditional on-premises IT, you had to buy servers to handle peak demand. Those servers sat idle the rest of the time, still generating costs. In the cloud, you scale up only during peaks and scale down otherwise, cutting costs.
Exam Key Points
"Borrow computing resources over the internet, pay only for what you use" -- Definition of cloud computing