AZ-104 domain - 24% of the exam

Deploy and Manage Azure Compute Resources

Deploy and Manage Azure Compute Resources is 24% of the AZ-104 exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

59
Practice questions
4
Objectives
45
Medium
14
Hard

Where people go wrong

  • Confusing which slot settings follow a swap

    The wrong answers move settings that stay put and pin settings that travel. Custom domain names are slot specific and stay in the same slot after a swap, as do scale settings, publishing endpoints and virtual network integration. Language framework version, the 32-bit versus 64-bit platform setting, WebSockets, handler mappings, public certificates and path mappings are swapped and follow the content into production. Treating a custom domain as swapped, or handler mappings as pinned, reverses the rule.

    See it in a sample question
  • Using scale out to change tier or isolate apps

    The wrong answers reach for more instances when the problem is the tier or the plan. Scaling out adds instances of the same tier, so it does not unlock deployment slots, which need the Standard, Premium or Isolated tier. Autoscale changes the instance count for the whole plan, and apps in one plan still share those VM instances. Isolation needs a separate App Service plan. The Free tier cannot scale out at all, because it runs on a shared instance.

    See it in a sample question
  • Attributing a container feature to the wrong service or tier

    Each wrong answer credits a container feature to the wrong service, tier or capability. Geo-replication is a Premium Azure Container Registry feature, so it is not added to Basic or Standard by attaching a region, and it is not a Container Apps capability. Webhooks, geo-replication and content trust do not rebuild images; registry tasks do when a base image is updated. Per-second billing and exact CPU and memory sizing belong to Azure Container Instances, and registry tiers belong to the registry.

    See it in a sample question
  • Assuming default encryption covers the temporary disk

    The wrong answers credit default Azure Storage encryption with covering the VM temporary disk, or offer an in-guest or storage-layer option where encryption must start on the VM host. Server-side encryption covers managed OS and data disks at rest, but the temporary disk is not a managed disk. Azure Disk Encryption works inside the guest with BitLocker or DM-Crypt, and double encryption adds a second layer at the storage level. Encryption at host starts on the server hosting the VM, so the temporary disk and caches are encrypted without using VM CPU.

    See it in a sample question
  • Treating orchestration mode as adjustable

    The wrong answers offer a setting or a rule that would change a scale set's orchestration mode after creation. No such setting exists: the mode is defined when the scale set is created and cannot be changed or updated later. A Uniform scale set that now needs standard IaaS VMs, or mixed Linux and Windows instances, must be replaced by a new Flexible one. An autoscale rule cannot switch the mode, and the mode does not decide which in-guest counters autoscale can read.

    See it in a sample question

Objectives in this domain

Sample questions from this domain

Free sampleDeploy and Manage Azure Compute Resourceshard

An app has a staging slot and a production slot. A custom domain name is bound to the production slot. After you swap staging into production, the operations team is surprised the custom domain did not move to the staging slot. Why did the custom domain stay with production?

  • ACustom domain names follow the content and swap along with app settings.
  • BCustom domain names swap only when the source slot is fully warmed up.
  • CCustom domain names swap only on plans in the Isolated tier.
  • DCustom domain names are slot specific and stay in the same slot after a swap. Correct
Custom domain names are slot specific and do not move during a slot swap. Custom domain names are in the group of settings that are not swapped, so they remain pinned to the slot they were configured on rather than following the content during a swap.

Why A is wrong: App settings can swap, but custom domain names are listed among the settings that are not swapped, so they do not follow the content.

Why B is wrong: Warm-up governs downtime during the swap, not whether a setting is classified as swappable; domains are not swapped regardless.

Why C is wrong: The swap classification of a setting does not vary by tier; custom domain names are not swapped on any tier.

Why D is correct: Correct. Custom domain names are in the group of settings that are not swapped, so they remain pinned to the slot they were configured on rather than following the content during a swap.

Free sampleDeploy and Manage Azure Compute Resourcesmedium

A web app on a Basic App Service plan needs the deployment slots feature, which Basic does not provide. You do not need more VM instances, only the feature. Which action gives the app access to deployment slots?

  • AScale out the plan by raising the number of running VM instances.
  • BRestart all running VM instances so the feature set is reloaded.
  • CRedeploy the application code so the missing feature is enabled.
  • DScale up the plan by changing it to a tier that includes the feature. Correct
Scale up means change the pricing tier to gain CPU, memory, disk, or features; scale out means add VM instances. You scale up by changing the pricing tier of the App Service plan, and scaling up is what grants extra features such as staging slots, custom domains and certificates, and autoscaling.

Why A is wrong: Scaling out adds instances of the same tier; the feature set is fixed by the tier, so adding instances does not unlock slots.

Why B is wrong: Restarting instances does not change the tier and the available features are determined by the pricing tier.

Why C is wrong: Scale settings need no code change or redeploy, and the feature depends on the tier rather than the deployed code.

Why D is correct: Correct. You scale up by changing the pricing tier of the App Service plan, and scaling up is what grants extra features such as staging slots, custom domains and certificates, and autoscaling.

Free sampleDeploy and Manage Azure Compute Resourcesmedium

Your team needs an Azure container registry that distributes the same container images across several geographies so each regional deployment pulls from network-close storage. Which registry tier provides the geo-replication feature required for this distribution?

  • AStandard
  • BBasic
  • CPremium Correct
  • DAny tier with an added replica
Geo-replication is a Premium-only feature among the Basic, Standard, and Premium registry tiers. The geo-replication feature of Premium registries supports advanced replication and container image distribution across geographies, so a Premium registry meets the cross-geography requirement.

Why A is wrong: Standard supports the common features such as webhooks and Entra authentication, but geo-replication is a Premium capability.

Why B is wrong: Basic offers the common features and the smallest storage and limits, and does not provide geo-replication.

Why C is correct: Correct. The geo-replication feature of Premium registries supports advanced replication and container image distribution across geographies, so a Premium registry meets the cross-geography requirement.

Why D is wrong: It is tempting to assume geo-replication is just a replica you can attach to any registry, but the feature is exclusive to Premium and is not available on Basic or Standard by adding a region.

Free sampleDeploy and Manage Azure Compute Resourcesmedium

Your security team requires that data flowing from a running VM to Azure Storage, including the VM temporary disk and the OS and data disk caches, is encrypted before it leaves the VM host, without consuming VM CPU. The VM has never had Azure Disk Encryption applied. Which option meets this requirement?

  • AEnable encryption at host on the virtual machine Correct
  • BEnable Azure Disk Encryption on the virtual machine guest
  • CRely on the default Azure Storage server-side encryption only
  • DApply double encryption at rest using platform-managed keys
Encryption at host encrypts the temporary disk and disk caches on the VM host with no CPU cost, giving end-to-end encryption. Encryption at host starts encryption on the Azure server hosting the VM, where the temporary disk and OS and data disk caches are stored, so all that data is encrypted at rest and flows encrypted to Storage. It does not use the VM CPU and does not impact performance.

Why A is correct: Correct. Encryption at host starts encryption on the Azure server hosting the VM, where the temporary disk and OS and data disk caches are stored, so all that data is encrypted at rest and flows encrypted to Storage.

Why B is wrong: Azure Disk Encryption uses BitLocker or DM-Crypt inside the guest and does not encrypt the temporary disk and disk caches end to end on the host the way encryption at host does.

Why C is wrong: Default server-side encryption protects managed OS and data disks at rest in Storage but does not cover the temporary disk, which is not a managed disk.

Why D is wrong: Double encryption adds a second encryption layer at the infrastructure storage layer but does not move encryption onto the VM host for the temporary disk and caches.

Free sampleDeploy and Manage Azure Compute Resourcesmedium

You created a Virtual Machine Scale Set in Uniform orchestration mode three weeks ago. A new requirement means you now need the scale set to run standard Azure IaaS VMs and mix Linux and Windows instances. Which action lets you meet this requirement?

  • ASwitch the existing scale set to Flexible orchestration through the Scaling settings blade.
  • BDeploy a new scale set, choosing Flexible orchestration at creation, then migrate the workload. Correct
  • CUpdate the scale set upgrade policy to Rolling so mixed instance types become allowed.
  • DAdd an autoscale rule that provisions standard IaaS virtual machines into the scale set.
Orchestration mode is fixed at creation. Choose Flexible up front if you need standard IaaS VMs or mixed OS instances. The orchestration mode is defined when the scale set is created and cannot be changed or updated later, so meeting the new requirement means creating a new scale set with Flexible orchestration, which manages standard Azure IaaS VMs and can mix Linux and Windows instances.

Why A is wrong: There is no setting that converts an existing scale set between orchestration modes after creation.

Why B is correct: Correct. The orchestration mode is defined when the scale set is created and cannot be changed or updated later, so meeting the new requirement means creating a new scale set with Flexible orchestration, which manages standard Azure IaaS VMs and can mix Linux and Windows instances.

Why C is wrong: The upgrade policy governs how instance updates roll out, not the orchestration mode or instance type mixing.

Why D is wrong: Autoscale changes instance count using the scaling profile and does not alter the orchestration mode.

Other domains in this exam

See also the AZ-104 cert hub, the study guide, and the cheat sheet.

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