AZ-900 domain - 29% of the exam

Describe Cloud Concepts

Describe Cloud Concepts is 29% of the AZ-900 exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

72
Practice questions
3
Objectives
24
Easy
48
Medium

Where people go wrong

  • Confusing hybrid cloud with multicloud

    The wrong options swap these two models whenever a scenario mentions more than one cloud. Multicloud means deliberately using two or more public cloud providers, with no private component. Hybrid cloud interconnects a private cloud and a public cloud, which lets a private cloud surge into public resources for temporary demand and lets a team choose per service which workloads stay private. Because hybrid places some data on shared public hardware, it cannot meet a requirement that data never sit alongside another tenant.

    See it in a sample question
  • Moving fixed responsibilities between customer and provider

    The wrong options move responsibility to whichever party seems closer to it. Some give stored data and identities to the provider because it hosts the service, and others give the physical datacentre, network and hosts to the customer because the customer pays for capacity. The shared responsibility model fixes both sets in every service model. Stored data, accounts and identities stay with the customer, even under SaaS, while the physical hosts, network and datacentre always stay with the provider.

    See it in a sample question
  • Matching operating system control to the wrong service model

    The wrong options pair a requirement with a model that removes or adds the wrong layer. PaaS and SaaS have the provider maintain the operating system, so neither suits a team that must control OS versions, patch schedules or bespoke dependencies. IaaS leaves operating system installation, patching and configuration with the customer, which suits rehosted virtual machines. A team that wants to stop patching operating systems and databases needs PaaS, because IaaS keeps that work and SaaS offers no place to run custom code.

    See it in a sample question
  • Mixing up vertical and horizontal scaling

    Scaling terms get attached to the wrong action. Vertical scaling changes the size of one resource by adding or removing CPU or RAM, and scaling up and scaling down are its two directions. Horizontal scaling changes the number of instances, with scaling out adding identical instances and scaling in removing them. The wrong options offer scaling out when one machine is enlarged, a vertical action when instances are added, or scaling in and scaling down when demand is rising. A machine already at its largest size can only scale out.

    See it in a sample question
  • Swapping cloud benefits that sound alike

    Benefit names get matched to the wrong definition. High availability is maximum uptime backed by service-level agreements. Reliability is recovering from failures and continuing to function, helped by spreading a workload across regions. Scalability adjusts resources to meet demand, and governance uses templates and auditing to keep resources within technical and regulatory standards. The wrong options pick an overlapping benefit, such as reliability for an SLA question, or high availability for a recover-from-failure question.

    See it in a sample question

Objectives in this domain

Sample questions from this domain

Free sampleDescribe Cloud Conceptsmedium

A government department must guarantee that its workloads never share underlying hardware with any other organisation's data. Which deployment model is designed to satisfy this specific requirement?

  • APublic cloud, because the provider isolates each customer logically across its shared pool of hardware
  • BHybrid cloud, because it lets sensitive workloads surge into the public tier during demand peaks
  • CPrivate cloud, because resources serve a single entity and its data is not collocated with other tenants Correct
  • DMulticloud, because spreading workloads across providers keeps any single tenant's data separated
A private cloud serves a single entity, so its data is never collocated with other tenants. A private cloud is consumed by a single entity using resources that belong to that organisation, so its data is not collocated with other tenants on shared hardware, which is precisely what a strict no-shared-infrastructure requirement demands.

Why A is wrong: A public cloud's resources are shared among many customers, so even with logical isolation the hardware itself is multi-tenant, which fails a no-shared-hardware mandate.

Why B is wrong: Hybrid interconnects public and private and can surge workloads outward, but bursting into shared public hardware is exactly what the no-collocation requirement forbids.

Why C is correct: Correct. A private cloud is consumed by a single entity using resources that belong to that organisation, so its data is not collocated with other tenants on shared hardware, which is precisely what a strict no-shared-infrastructure requirement demands.

Why D is wrong: Multicloud uses several public providers whose infrastructure is still multi-tenant, so spreading across them does not deliver dedicated single-entity hardware.

Free sampleDescribe Cloud Conceptseasy

A new cloud customer wants to know which responsibility they retain no matter whether they adopt IaaS, PaaS, or SaaS. Which set always stays with the customer?

  • ATheir stored data and their accounts and identities Correct
  • BThe physical datacentre and the racks within it
  • CThe physical network that links the servers
  • DThe physical hosts that run the workloads
Your data and your identities are always your responsibility in the cloud, whatever the service model. Across the shared responsibility model, data or information stored in the cloud and the accounts and identities of people, services, and devices are listed as always the customer's, irrespective of whether the service is IaaS, PaaS, or SaaS, because only the customer can govern who and what they trust.

Why A is correct: Correct. Across the shared responsibility model, data or information stored in the cloud and the accounts and identities of people, services, and devices are listed as always the customer's, irrespective of whether the service is IaaS, PaaS, or SaaS, because only the customer can govern who and what they trust.

Why B is wrong: The physical datacentre is real and tempting because the customer pays for capacity, but the model lists it as always the provider's responsibility, never the customer's.

Why C is wrong: The physical network is a genuine shared-responsibility item, yet the model assigns the physical network to the provider in every service model, not the customer.

Why D is wrong: Physical hosts feel customer-adjacent because workloads run on them, but the model places the physical hosts permanently with the provider regardless of service type.

Free sampleDescribe Cloud Conceptsmedium

An organisation plans to move its existing on-premises virtual machine workloads to the cloud with minimal redesign, recreating a datacenter-like environment while retaining full control over operating system installation, patching, and configuration. Which cloud service model best meets these requirements?

  • APlatform as a service (PaaS), using the managed application platform components
  • BSoftware as a service (SaaS), subscribing to a finished hosted application
  • CA managed database service that the provider patches and operates on the customer's behalf
  • DInfrastructure as a service (IaaS), renting the underlying compute and storage hardware Correct
IaaS gives maximum control and OS-level access, making it the natural fit for lift-and-shift migration of existing virtual machines. Under IaaS the provider maintains only the physical hardware, networking, and security, while the customer manages operating systems, configuration, and patching, so existing virtual machines can be rehosted almost unchanged with full operational control retained.

Why A is wrong: PaaS is a managed development platform where the provider maintains the operating systems and middleware, which is attractive for new development but removes the OS-level control the organisation wants to preserve when migrating its existing virtual machines unchanged.

Why B is wrong: SaaS provides a fully developed application that the customer simply uses, so it is the easiest model but offers nothing to host arbitrary existing virtual machine workloads and no control over the operating system at all.

Why C is wrong: A managed database is a single platform service the provider patches and operates, which is tempting because migration often involves databases, but it cannot host general virtual machine workloads and surrenders the OS control the lift-and-shift requires.

Why D is correct: Correct. Under IaaS the provider maintains only the physical hardware, networking, and security, while the customer manages operating systems, configuration, and patching, so existing virtual machines can be rehosted almost unchanged with full operational control retained.

Free sampleDescribe Cloud Conceptsmedium

An engineer keeps a single virtual machine in place but gives it more CPU and additional RAM so the same instance can handle a heavier workload. Which kind of scaling has the engineer performed?

  • AHorizontal scaling, by adjusting the number of running instances
  • BScaling out, by adding further virtual machines to the pool
  • CLoad balancing, by spreading requests across many servers
  • DVertical scaling, by increasing the size of the one instance Correct
Vertical scaling changes the size or power of a single existing resource by adding CPU or RAM. Vertical scaling means increasing or decreasing the capabilities of a resource, and scaling up specifically adds more CPU or RAM to an existing virtual machine, which is what happens when one instance is made more powerful without changing the instance count.

Why A is wrong: Horizontal scaling is a real approach but is wrong here because it changes the count of instances by adding or removing machines, whereas this engineer kept one machine and enlarged it.

Why B is wrong: Scaling out is genuine and tempting because it also raises capacity, but it does so by adding more instances rather than increasing the size of the single existing virtual machine.

Why C is wrong: Load balancing is a real performance technique that distracts because it also handles heavier load, but it redirects traffic across multiple resources rather than enlarging a single virtual machine.

Why D is correct: Correct. Vertical scaling means increasing or decreasing the capabilities of a resource, and scaling up specifically adds more CPU or RAM to an existing virtual machine, which is what happens when one instance is made more powerful without changing the instance count.

Free sampleDescribe Cloud Conceptseasy

A finance team asks which cloud benefit keeps their systems reachable through disruptions, with uptime commitments written into service-level agreements (SLAs). Which benefit are they describing?

  • AReliability, the capacity to recover from failures and keep functioning
  • BScalability, the ability to adjust resources to meet shifting demand
  • CHigh availability, ensuring maximum uptime that is backed by SLAs Correct
  • DCost predictability, forecasting future spend from real usage
High availability means maximum uptime regardless of disruptions, with the guarantees written into service-level agreements. High availability is the benefit specifically defined as ensuring maximum availability regardless of disruptions, and the uptime levels a service promises are formalised as guarantees within its service-level agreements, which is precisely what the finance team is asking about.

Why A is wrong: Reliability is a genuine Azure benefit and tempts candidates because it also concerns staying operational, but it describes recovering from and resisting failures across regions rather than the uptime guarantees expressed in SLAs.

Why B is wrong: Scalability is a real benefit but tempts candidates by overlap with growth, yet it concerns adjusting resource capacity to demand, not guaranteeing uptime through SLAs.

Why C is correct: Correct. High availability is the benefit specifically defined as ensuring maximum availability regardless of disruptions, and the uptime levels a service promises are formalised as guarantees within its service-level agreements, which is precisely what the finance team is asking about.

Why D is wrong: Cost predictability is a legitimate benefit that appeals because budgeting feels foundational, but it addresses forecasting spend from monitored usage, not the availability commitments in an SLA.

Other domains in this exam

See also the AZ-900 cert hub, the study guide, and the cheat sheet.

Examworthy is not affiliated with or endorsed by Microsoft. Original, blueprint-aligned practice material only.