Study Azure Virtual Desktop as a service, not a collection of settings
Microsoft AZ-140 assesses an administrator’s ability to plan, implement, secure, deliver and maintain an Azure Virtual Desktop environment. A useful study sequence should follow the dependencies of a working service: user requirements lead to host-pool and network design; image and profile decisions influence session reliability; identity and security determine who can connect; applications and user settings determine usefulness; and monitoring and recovery reveal whether the deployment can be operated. Starting from the user’s experience makes the large blueprint manageable without reducing it to a checklist of unrelated Azure features.
Microsoft’s July 20, 2026 exam objectives allocate 40–45% to Azure Virtual Desktop infrastructure, 15–20% to identity and security, 20–25% to user environments and apps, and 10–15% to monitoring and maintenance. These approximate weights are a coverage guide, not a quota of exact questions. The active Microsoft AZ-140 exam is associated with Microsoft Certified: Azure Virtual Desktop Specialty. Verify current exam scheduling and qualification rules with Microsoft, and use the percentage split to prevent studying only host pools while overlooking access, profiles and recovery.
Before building anything, describe three user groups in a hypothetical organization: occasional task workers, knowledge workers who need persistent personalization, and a small group using demanding applications. Estimate concurrency, latency sensitivity, application compatibility, geography and acceptable interruption. This planning scenario will anchor later decisions about pooled versus personal desktops, VM size, image strategy, identity, storage and autoscaling. Record each requirement and its verification method so the same case can be revisited throughout preparation.
Begin with the client path, networking and storage requirements
Start with the path a user takes to reach an Azure Virtual Desktop session. Draw the client, authentication flow, host pool, session host, virtual network, name resolution and route toward required applications or profile storage. Review network capacity, DNS, NSGs, user-defined routes, Azure Firewall and private connectivity where appropriate. Microsoft’s current skills list specifically includes RDP Shortpath, RDP Multipath, QoS and Azure Private Link. The point of a connection diagram is to explain which dependency can cause a session to fail or perform poorly, not to add every networking product to one topology.
Use a small troubleshooting exercise to distinguish an authentication failure from a transport failure and from an application that cannot be reached after the session starts. Examine the expected session path, relevant identity signals, session-host reachability and network performance counters. A blocked policy can look like an unavailable desktop; a high-latency path can look like a slow VM. Learning these boundaries before deploying a complex host pool avoids diagnosing every user complaint by restarting virtual machines.
Plan user-data storage at the same time as host-pool architecture. FSLogix Profile Containers, ODFC containers and Cloud Cache have different purposes; the storage choice must match user persistence, performance, identity and recovery requirements. Compare Azure Files and Azure NetApp Files where applicable, including connectivity and permissions. A pooled host should be replaceable without losing user state. The dependency is easy to remember when you trace how a user profile is mounted at sign-in and what happens when the backing share is slow or unavailable.
Build host pools with an intentional image and capacity design
Create a small host-pool design from the user personas. Decide whether pooled or personal desktops fit, whether the chosen operating system and licensing support the scenario, how many session hosts are required at normal and peak load, and how users will be assigned. Define the subscription, resource group and management boundaries before creating anything. Then implement a disposable host pool through the portal or an approved template, and inspect registration, health and the configuration that determines where a user session is placed. A successful VM deployment is not the same as a registered, usable Azure Virtual Desktop session host.
Review supported automation using PowerShell, Azure CLI, ARM templates or Bicep. Do not memorize every syntax variation: understand what inputs create host pools and hosts, which parameters vary by environment, and which state must be verified after deployment. A reproducible configuration is easier to audit and rebuild than a manual series of portal operations with undocumented defaults. If you cannot run a full lab, write a deployment manifest and explain what a safe validation would look like without claiming the deployment occurred.
Next build an image lifecycle rather than configuring each host by hand. Compare standard and customized images, image storage in Azure Compute Gallery, updates, image versioning and provisioning new session hosts. Practice an approved maintenance change using drain mode and a replacement host, preserving active sessions as required by the service policy. The value of a consistent image becomes visible when a failed host can be replaced without recreating its application configuration or compromising a user’s profile.
Make identity and security part of the connection design
Select an identity scenario appropriate to the hypothetical organization: Microsoft Entra ID, Active Directory Domain Services, or Microsoft Entra Domain Services where supported. Know the host requirements and trade-offs rather than treating identity choices as aliases. Review Azure RBAC for management, role assignments, groups and the rights a user needs inside a session host. Then examine authentication to Azure Virtual Desktop, Microsoft Entra single sign-on, multifactor and passwordless options, and Conditional Access policies controlling who may sign in under which conditions.
Create a troubleshooting matrix separating account eligibility, host registration, Conditional Access, role assignment and Windows permissions. A user may successfully authenticate to Microsoft Entra and still not receive the desired desktop because the application group assignment is missing. Conversely, the host can be healthy while policy blocks a connection. Use a limited, permitted test identity to verify one intended access decision and one denied condition; do not weaken the organization’s security controls simply to make a test pass.
Study session-host protection as an operating requirement. Microsoft’s current outline includes Defender for Cloud, Defender Antivirus, Defender for Endpoint, NSGs, Azure Firewall, Bastion or just-in-time administration, App Control for Business, Controlled Folder Access, Trusted Launch and confidential VM considerations. These technologies operate at different layers. Understand which protects administrative access, which detects endpoint threats, which filters traffic and which helps harden the platform. Conditional Access belongs in this wider identity and session-security story, not as a substitute for host hardening.
Deliver applications and a dependable user experience
Once an authorized desktop connection works, study what the user sees. Application groups, workspaces and assignments determine whether someone receives a full desktop or a published RemoteApp. Configure or model a separate application group for a small test audience and verify the resource appears to the intended users. Review host-pool RDP properties, personal desktop assignment, session timeouts and Start VM on Connect. A setting that saves compute cost can still violate a service requirement if users wait too long for the desktop to become available.
Move on to client and device behavior. The July 2026 blueprint includes choosing and troubleshooting Azure Virtual Desktop clients, device and multimedia redirection, Universal Print, Microsoft Intune or Group Policy settings, and Remote Desktop Protocol properties. A failed printer, microphone or video call may be a client or redirection problem rather than a host-size problem. Practice following the relevant policy and client dependency before changing VM resources. Record the expected user experience for each change so the outcome can be tested.
Finally, review application deployment on multisession hosts: Microsoft 365 Apps, OneDrive, Microsoft Teams, browsers and dynamic application delivery through App attach. Image updates, package compatibility and user profiles interact, so an app that works during installation may still fail for a signed-in user after a new host image is published. Create an update test covering application registration, profile persistence, user assignment, client behavior and rollback. Keep this article’s purpose a study sequence; a separate Azure Virtual Desktop architecture discussion can deepen the placement and topology decisions.
Establish monitoring baselines before optimizing capacity
Monitoring becomes meaningful only after you know what healthy performance looks like. Collect baseline session count, sign-in duration, host CPU and memory pressure, network latency, profile-mount time and application response in a permitted test environment. Review Azure Monitor, Log Analytics and Azure Virtual Desktop Insights as ways to collect or interpret operational data. A complaint that the desktop is slow could originate in profile storage, a congested network, overloaded session hosts, client redirection or an application. Compare observations before taking corrective action.
Study autoscaling as a capacity and user-experience decision. Model a workday with morning demand, a peak period, disconnected sessions, drain mode and a quieter evening. Estimate the effect of VM size, session density and minimum running hosts on both cost and responsiveness. The cheapest configuration at midnight may create an unacceptable morning sign-in delay. A credible optimization proposal includes thresholds, monitoring evidence and safeguards that protect active work, not just a calculation of idle virtual machines.
Rehearse an operations incident: an autoscaling rule reduces available capacity while several users reconnect. Determine which diagnostic data would reveal the problem, how you would restore sufficient capacity, and what alert should have warned earlier. A second incident might involve a slow FSLogix share despite healthy session-host CPU. These examples show how monitoring and scaling depend on the user path learned earlier and why application-level evidence is needed alongside infrastructure metrics.
Prepare updates, backup and recovery as different failure problems
An Azure Virtual Desktop recovery plan distinguishes replaceable pooled hosts from user-state storage, personal desktops, images, configuration and service dependencies. A backup or snapshot of the wrong component may not restore a useful desktop experience. Review recovery of FSLogix profiles, personal desktop data, images and the configuration required to recreate session hosts. Document permissions and identity dependencies, not only storage copies. Decide which components need point-in-time recovery and which can be rebuilt from tested automation.
Use a maintenance rehearsal to demonstrate why host-image updates require operational control. Place a test host into drain mode, handle current sessions according to policy, provision a replacement from a new image, confirm registration and user access, and verify that profiles and applications work. If the new image creates a regression, explain how the rollout would be reversed. This is a stronger practical check than merely knowing how to create an image in the portal.
For multi-region disaster recovery, map networking, DNS, identity, profile storage, images, application packages, licensing and monitoring before drawing a second set of VMs. A secondary region with hosts but no reachable profiles or identity path may not satisfy recovery objectives. Build a tabletop scenario with a regional outage, identify required recovery-time and recovery-point targets, then state how operators would verify that users can work again. Microsoft’s outline explicitly includes updates, backups, disaster recovery and multi-region planning, making these service dependencies part of AZ-140 readiness.
Finish with two integrated cases and an evidence-based revision
End by tracing an ordinary user login from client through authentication, host assignment, network, profile mount, application launch and active session monitoring. Then break one dependency in a controlled scenario: remove an application assignment, block a network path, introduce a profile permission error or make a host unavailable. Identify the first diagnostic signal that narrows the problem and the smallest safe action that restores service. If you are learning with a simulator or design exercise, state what you would observe rather than inventing logs or production test results.
Use a second scenario for performance and operations. Users can connect, but morning sign-ins take too long and interactive applications lag. Compare capacity, autoscaling, session density, profile storage, network path and client redirection. Decide which measurements would distinguish the causes and what change you would test first. The objective is to move from a broad symptom to an evidence-backed recommendation rather than guessing that more VMs will fix every problem.
Make the final review adaptive. For the four published objective groups—40–45% infrastructure, 15–20% identity and security, 20–25% user environments and applications, and 10–15% operations—identify one design decision, one configuration or tabletop exercise, one failure and one recovery action. Study weak dependencies again with a different example. The strongest preparation result is being able to explain the Azure Virtual Desktop service lifecycle and diagnose its operational failures, not completing a fixed set of disconnected study tips.