Azure Virtual Machines: the AZ-900 essentials

One service the exam probes from every angle — the IaaS bargain, VM sizes, the three disk types, and how you keep a VM up. AZ-900, 2026 edition.

Here are 7 typical Azure Virtual Machines questions in the exam's own style — free, each answer explained, no sign-up.

Straight to the 7 free questions ↓

Azure Virtual Machines is the one Azure Compute service the AZ-900 keeps circling back to, because a single VM lets the exam test the IaaS model, sizing, storage, and uptime all at once. Learn the parts labelled in the diagram above and this whole cluster of questions turns into free points.

A VM is IaaS — the control-for-effort bargain

A Virtual Machine is Infrastructure as a Service. Microsoft runs the datacentre, the physical host, and the hypervisor; everything from the guest operating system upward is yours. You patch the OS, you harden it, you install what you like. That freedom is the whole point of a VM — and the maintenance it brings is its price. Any AZ-900 scenario where you are patching or configuring the OS is pointing at IaaS.

Sizes and series — match the shape, do not learn the list

Azure offers VM sizes grouped into series (families) tuned for different resource balances. You do not learn every size — you just recognise the idea:

General purpose — a balanced CPU-to-memory ratio for most workloads.
Compute-optimized — a high CPU-to-memory ratio for CPU-bound jobs.
Memory-optimized — lots of RAM for in-memory and data workloads.

Match the workload’s shape to the family and move on.

Three disks, three jobs

A VM attaches storage in three roles, and the exam loves the distinction:

Managed OS disk — holds the operating system; persists.
Managed data disk — holds your application data; persists.
Temporary disk — local scratch storage that can be wiped on deallocation or host maintenance.

The rule to know cold: anything you would miss goes on a managed OS or data disk; the temporary disk is scratch only.

Keeping a VM up — sets, zones, and the SLA

Two uptime constructs are easy to confuse:

Availability Set — spreads VMs across fault and update domains inside one datacentre, surviving a rack failure or a host patch.
Availability Zones — spread VMs across physically separate datacentres in a region, surviving a whole-datacentre outage.

Because redundancy differs, the VM service-level agreement (SLA) depends on how you deploy: a single VM, an Availability Set, and zone-spread VMs each carry a different, rising availability commitment. Learn the shape, not the exact percentages.

Scale Sets and Spot VMs

Two more names round out the topic. A Virtual Machine Scale Set manages a group of identical VMs and adds or removes instances automatically — the elastic web tier. A Spot VM sells Azure’s spare capacity at a deep discount in exchange for being interruptible, perfect for restartable, cost-sensitive batch work.

The real exam mixes multiple-choice, true/false, and drag-and-drop formats; the seven questions below are standard multiple-choice, and every explanation carries a “why not the others” so the wrong options teach you as much as the right one.

A labelled diagram of a Microsoft Azure Virtual Machine and its parts. A central VM box carries a size and series label (for example a general-purpose D-series) and shows three attached disks distinguished by role: a managed OS disk that persists and holds the operating system, one or more managed data disks that persist and hold application data, and a temporary disk marked as scratch storage that can be lost on deallocation or host maintenance and must never hold data you need to keep. A network interface card connects the VM to a virtual network. Below, a note contrasts two uptime constructs: an Availability Set, which spreads VMs across fault and update domains inside a single datacentre, versus Availability Zones, which spread VMs across physically separate datacentres in one region — with a reminder that the VM service-level agreement depends on which redundancy you choose. A side caption notes that a Virtual Machine Scale Set manages a group of identical VMs for autoscale, and that a Spot VM offers cheap, interruptible capacity.

7 free AZ-900 practice questions

Answers and explanations — no email wall

AZ-900 Question 1 of 7

Your team deploys an Azure Virtual Machine and installs its operating system updates itself, schedules its own patch windows, and hardens the OS. Which cloud service model does this reflect, and who is responsible for patching the guest OS?

Answer: A — Infrastructure as a Service (IaaS) — you are responsible for the guest OS, including patching.

A Virtual Machine is the classic Infrastructure as a Service offering: Microsoft runs the physical host, hypervisor, and datacentre, but everything from the guest operating system upward is yours — patching, configuration, and installed software included. Why not the others: with PaaS the platform patches the OS for you, which is exactly the burden a VM does not remove. SaaS delivers finished software with no infrastructure exposed. Function as a Service (serverless) hides the OS, so there is nothing for you to patch. Pro tip: if the scenario has you patching or configuring the OS, it is IaaS and the responsibility is yours.

AZ-900 Question 2 of 7

An analytics workload is CPU-bound and needs a high ratio of virtual CPUs to memory, while a separate in-memory caching workload needs the opposite. In Azure, how do you match each to the right hardware?

Answer: A — Choose a compute-optimized VM size for the CPU-bound job and a memory-optimized VM size for the caching job.

Azure groups VM sizes into series (families) tuned for different balances of resources: general purpose for balanced workloads, compute-optimized for a high CPU-to-memory ratio, and memory-optimized for large in-memory data. You pick the family that fits the workload's shape. Why not the others: general purpose is a balanced middle, not the best fit for either extreme. Azure Policy governs what is allowed — it does not reassign hardware by workload. Spot VMs and Availability Sets address cost and uptime, not the CPU-versus-memory balance. Pro tip: you do not learn every size — recognise the idea that families exist (general purpose, compute-optimized, memory-optimized) and match the shape.

AZ-900 Question 3 of 7

You need storage on an Azure VM that survives a reboot and a deallocation, and you are told never to place important files on the drive Azure provisions automatically for scratch space. Which disk holds your data safely?

Answer: A — A managed data disk, which persists independently of the VM's power state.

An Azure VM has three disk roles: a managed OS disk (holds the operating system, persists), one or more managed data disks (hold your application data, persist), and a temporary disk (local scratch storage that can be wiped on deallocation or host maintenance). Persistent data belongs on a managed data disk. Why not the others: the temporary disk being fast or free does not make it safe — its whole caveat is that it is not durable. And persistent managed disks plainly survive deallocation, so copying to on-premises is unnecessary. Pro tip: temporary disk = scratch only; anything you would miss goes on a managed OS or data disk.

AZ-900 Question 4 of 7

You must protect a VM-based application against the failure of a single physical rack or a host update inside one datacentre, without spanning multiple datacentres. Which construct is designed for this?

Answer: A — An Availability Set, which spreads VMs across fault and update domains within a datacentre.

An Availability Set places VMs across multiple fault domains (separate racks/power/network) and update domains (patched at different times) inside one datacentre, protecting against localised hardware failure and host maintenance. Why not the others: an Availability Zone is not a rack — it is a physically separate datacentre in a region, used to survive a whole-datacentre outage. A Spot VM is cheap, interruptible capacity and guarantees nothing about racks. A resource group is a management and billing container; it provides no availability by itself. Pro tip: Availability Set = inside one datacentre (racks and patch waves); Availability Zone = across datacentres in a region.

AZ-900 Question 5 of 7

A retail site's web tier must automatically add identical VMs during a traffic spike and remove them afterwards, all from one configuration. Which Azure feature provides this?

Answer: A — A Virtual Machine Scale Set, which manages a group of identical VMs with autoscale.

Virtual Machine Scale Sets let you deploy and manage a group of identical, load-balanced VMs and scale the count in or out automatically in response to demand or a schedule — exactly the elastic web tier described. Why not the others: resizing one VM changes its power, not its count, and cannot scale beyond a single machine's limits. An Availability Set improves uptime but does not add or remove instances on load. Azure Advisor only recommends improvements; it does not provision compute. Pro tip: 'many identical VMs that grow and shrink automatically' is the Scale Sets giveaway.

AZ-900 Question 6 of 7

A batch rendering job can be paused and restarted at any time and simply needs the cheapest possible compute. It is fine if Azure reclaims the capacity at short notice. Which VM purchasing option fits best?

Answer: A — A Spot VM, which offers deeply discounted, interruptible capacity.

Spot VMs sell Azure's spare capacity at a large discount, with the trade-off that Microsoft can evict them when it needs the capacity back. For interruptible, restartable work like batch rendering, that trade-off is ideal. Why not the others: a standard on-demand VM costs more and its resilience is unneeded here. Memory-optimized is about the resource balance, not price or interruptibility. A reservation lowers cost by committing for one or three years, but it is for steady, long-running workloads — not short bursts you are happy to have interrupted. Pro tip: interruption-tolerant plus cost-sensitive equals Spot.

AZ-900 Question 7 of 7

Two teams ask what uptime guarantee their VMs carry. One runs a single VM with a standard managed disk; the other spreads VMs across Availability Zones. What determines the service-level agreement (SLA) each receives?

Answer: A — The SLA depends on the redundancy configuration — more resilient deployments carry a higher availability commitment.

The financially backed availability commitment for VMs varies with how you deploy them: a single VM on premium/managed disks, VMs in an Availability Set, and VMs spread across Availability Zones each carry a different SLA, with more redundancy earning a higher one. Why not the others: it is not one fixed number for all deployments, and it is not driven by VM size. VMs do have an SLA — the point is that you raise it by adding redundancy. Pro tip: remember the shape, not the exact percentages — availability rises as you move from one VM, to an Availability Set, to Availability Zones.

That is exactly how every question in the course works — answer, explanation, why-not. The real set continues in the practice player: 10 free questions, no sign-up.

Continue with the 10 free questions →

Those 7 questions were the start.

The exam does not test whether you recognise a term — it tests whether you can rule out three plausible answers under time pressure. That is what the explanations above are for, and there are 300 more questions built exactly like them.

Collecting questions yourself

Scattered across forums, of unknown age
Answer keys without reasoning
No idea which domain you are weak in

Practising with a system

300 questions in 6 full tests, AZ-900 (2026)
Every option explained — including the wrong ones
Readiness per exam domain, and drills for your weakest

Start free — 10 questions

Straight into the player. No account, no email.

Frequently asked questions

Who is responsible for patching an Azure Virtual Machine?

You are. A Virtual Machine is Infrastructure as a Service, so Microsoft manages the physical host, hypervisor, and datacentre, but everything from the guest operating system upward — OS patching, configuration, and installed software — is your responsibility. That control is the reason to choose a VM, and the ongoing maintenance is its cost. If you would rather Microsoft patched the OS for you, a Platform as a Service option such as Azure App Service is the trade to consider.

What is the difference between an Availability Set and an Availability Zone?

An Availability Set spreads your VMs across fault domains and update domains inside a single datacentre, protecting against a rack-level hardware failure or a host update. An Availability Zone is a physically separate datacentre within a region, so spreading VMs across zones protects against an entire datacentre going down. In short: a set guards against failures inside one building; zones guard against losing a whole building. Deployments spread across zones carry a higher availability commitment.

Are these real exam questions?

No. They are our own questions, written in the style and difficulty of the AZ-900 — never copied from any question bank or the real exam. Reproducing live exam items violates Microsoft's certification agreement and can cost a candidate their certification, and a crammed answer teaches you nothing about the one you have not seen.

Updated for AZ-900 (July 2026). The sample questions above are our own work in the style of the exam — not real exam items. The exam itself is set and marked by the certification body.