AZ-104 Deploy and Manage Azure Compute Practice Questions

Question 1

You have an application that writes heavily to Azure-managed disks and requires the highest consistent IOPS and lowest latency. Which disk type should you choose?

Accepted Answer

Ultra Disk. Azure Ultra Disk offers the highest performance among Azure managed disks, with up to 160,000 IOPS per disk and sub-millisecond latency, exceeding Premium SSD v2's 80,000 IOPS. Therefore, Ultra Disk is the correct choice for applications requiring the highest consistent IOPS and lowest latency.

Answer

Standard HDD

Answer

Standard SSD

Answer

Premium SSD v2

Question 2

You need to run a scheduled script on an Azure virtual machine every time the VM starts. Which feature should you use?

Accepted Answer

A Custom Script Extension. To run a script every time an Azure VM starts, use the Custom Script Extension to deploy a script that creates a scheduled task configured to run at system startup. The Custom Script Extension itself executes only during provisioning or when manually invoked; it does not rerun automatically on each boot. Therefore, B is correct because it is the only option that allows executing custom code, and you can use it to set up the startup trigger.

Answer

Boot diagnostics

Answer

Azure Policy

Answer

An availability set

Question 3

You are deploying a Windows Server VM for an internal app. The VM must support Secure Boot and vTPM later, its OS disk must survive host moves, and the team wants the lowest-cost managed disk tier that still behaves like a normal writable OS disk. Which two choices should you make? Select two.

Accepted Answer

Use a Generation 2 Windows Server Marketplace image.. Secure Boot and vTPM are only supported on Generation 2 VMs in Azure, which use UEFI firmware and GPT-partitioned OS disks. A Generation 2 Windows Server Marketplace image provides the necessary firmware and partitioning. Additionally, Azure Trusted Launch (which enables Secure Boot and vTPM) requires a managed OS disk and supports only Premium SSD and Standard SSD tiers, not Standard HDD. Therefore, Standard SSD is the lowest-cost managed disk tier that supports vTPM while remaining a durable, writable OS disk. Ephemeral disks do not survive host moves and unmanaged VHDs are not managed disks, so they are incorrect.

Answer

Use an ephemeral OS disk to reduce storage cost.

Answer

Use an unmanaged VHD stored in a storage account.

Answer

Use a Generation 1 image and enable Secure Boot after deployment.

Question 4

An application runs on two identical VMs in a region that does not support availability zones. The app must keep running through planned maintenance and a single hardware fault, and the team does not want to add a second region. Which two deployment choices are appropriate? Select two.

Accepted Answer

Place the VMs in an availability set.. A is correct because an availability set distributes VMs across multiple fault domains (up to 3) and update domains (up to 20) within the same datacenter. This ensures that during planned maintenance (update domain isolation) or a hardware fault (fault domain isolation), at least one VM remains available, meeting the requirement without needing availability zones or a second region.

Answer

Put both VMs in the same fault domain to simplify patching.

Answer

Deploy the workload in availability zones anyway.

Answer

Use a second Azure region for the primary failover design.

Question 5

A reporting application must run on an Azure VM with at least 8 vCPUs and 64 GiB of RAM. The team also wants headroom for short spikes without falling below the requirement. Which two VM sizes meet or exceed the requirement? Select two.

Accepted Answer

Standard_E8s_v5. Standard_E8s_v5 is correct because it provides 8 vCPUs and 64 GiB of RAM, exactly meeting the stated minimum requirement. Standard_M8ms is also correct because it offers 8 vCPUs and 218.75 GiB of RAM, which exceeds both the vCPU and memory requirements and provides substantial headroom for short spikes. Standard_D8s_v5 does not belong because it has 8 vCPUs but only 32 GiB of RAM, falling short of the 64 GiB requirement. Standard_F8s_v2 is incorrect because it has 8 vCPUs but only 16 GiB of RAM. Standard_B8ms is incorrect because it provides 8 vCPUs but only 32 GiB of RAM, so it also fails the memory requirement.

Answer

Standard_D8s_v5

Answer

Standard_F8s_v2

Answer

Standard_B8ms

Question 6

A build pipeline starts a Linux container once per request. Each run lasts about 12 minutes, never needs inbound connections, and should not leave an always-on server running afterward. Which two configuration choices best fit Azure Container Instances? Select two.

Accepted Answer

Use Azure Container Instances for the workload.. Option A is correct because Azure Container Instances (ACI) is the serverless container service designed for short-lived, on-demand container runs without managing an always-on host, which matches the 12-minute per-request build job. Option B is correct because setting the container group's restartPolicy to Never ensures the container runs once and is not restarted after completion, so nothing keeps running after the build finishes. Option C is wrong because App Service plans with deployment slots are for continuously hosted web apps, not one-shot container jobs. Option D is wrong because a VM scale set hosts containers on persistent VMs, leaving infrastructure running and requiring management. Option E is wrong because availability sets apply to VMs for host fault/update protection, not to serverless ACI container groups.

Answer

Create an App Service plan with deployment slots.

Answer

Use a virtual machine scale set to host the container.

Answer

Place the workload in an availability set for host protection.

Question 7

A reporting application needs an Azure VM with at least 8 vCPUs and 64 GiB of RAM. The workload is memory-heavy, and the team wants a reasonable starting point. Which VM family should the administrator choose?

Accepted Answer

E-series. The E-series is Azure's memory-optimized VM family, designed for workloads that require a high memory-to-CPU ratio. For a memory-heavy reporting application, the E-series (e.g., E8s_v3 with 8 vCPUs and 64 GiB RAM) provides the needed memory efficiently and is a reasonable starting point. While a D-series VM like D16s_v3 can also meet the minimum specs, it does so with double the vCPUs, making E-series a more cost-effective and appropriate choice for memory-intensive workloads.

Answer

B-series

Answer

D-series

Answer

F-series

Question 8

A platform team maintains a hardened Windows VM with IIS, an agent, and local configuration files. They need to deploy 40 identical VMs in two regions, preserve version history, and roll back quickly if a new image causes problems. Which two actions should they take? Select two.

Accepted Answer

Generalize the source VM before capture by removing machine-specific state.. Option A is correct because generalizing the source VM with Sysprep (or waagent -deprovision for Linux) removes machine-specific state such as SIDs, computer name, and user accounts, which is required before capturing a Windows image that will be used to provision multiple identical VMs. Option B is correct because Azure Compute Gallery (formerly Shared Image Gallery) stores image definitions and versioned image versions, providing the version history and the ability to roll back to a previous image version when a new image causes problems, while also supporting multi-region replication for the two regions. Option C is not correct because availability sets only provide fault and update domain separation for VMs within a single datacenter, not image versioning or multi-region deployment. Option D is not correct because a user-assigned managed identity provides authentication to Azure resources and has no role in image capture, versioning, or rollback. Option E is not correct because boot diagnostics only capture serial console and screenshot output for troubleshooting VM boot issues; they do not preserve image version history or enable rollback.

Answer

Place the target virtual machines in an availability set.

Answer

Attach a user-assigned managed identity to the source VM.

Answer

Enable boot diagnostics on the target virtual machines.

Question 9

A support desk needs to reset the local administrator password on specific virtual machines by using the VMAccess extension and restart those VMs. They must not be able to resize the machines, change networking, or manage disks. What should the administrator create?

Accepted Answer

Create a custom role with only the required VM actions and assign it at the virtual machine scope. The support desk needs only specific actions to reset the local admin password via the VMAccess extension (Microsoft.Compute/virtualMachines/extensions/write) and to restart VMs (Microsoft.Compute/virtualMachines/restart/action), without permissions to resize, change networking, or manage disks. A custom role at the VM scope allows granting precisely these required actions while excluding broader management actions. This follows the principle of least privilege and prevents unintended modifications to other resources.

Answer

Assign the Virtual Machine Contributor built-in role at the subscription scope

Answer

Assign the Contributor role at the resource group scope

Answer

Use an Azure Policy assignment to allow the VMAccess extension

Question 10

A business-critical application runs in a region that does not support availability zones. It uses two Azure VMs and must survive planned maintenance and a single host failure, but it does not need automatic scale-out. Which two actions should the administrator take? Select two.

Accepted Answer

Place both VMs in the same availability set. An availability set provides redundancy within a single region that does not support availability zones, protecting against both planned maintenance (via update domains) and host failures (via fault domains). By placing both VMs in the same availability set, Azure ensures they are distributed across multiple fault domains (up to 3) and update domains (up to 20), so a single hardware failure or planned maintenance event does not affect both VMs simultaneously.

Answer

Deploy the VMs so Azure distributes them across fault and update domains within that set

Answer

Deploy the VMs in separate availability zones

Answer

Convert the workload to a single larger VM

Answer

Put the VMs in a proximity placement group

Deploy and Manage Azure Compute

AZ-104 Deploy and Manage Azure Compute — Key Topics

Where candidates lose marks on Deploy and Manage Azure Compute