Assess legacy VDI applications one workload at a time, against the people who use them and the cloud desktop environment you plan to run. Inventory each application’s operating-system and session requirements, installation and licensing behavior, user data, peripherals, network dependencies, and business criticality; then test real workflows in a proof of concept. The result should be a per-application decision to migrate, remediate, replace, isolate, retain, or retire—not an assumption that an existing desktop image can simply be copied to the cloud.
Why is moving VDI to the cloud more than copying a desktop image?
A virtual desktop depends on more than its operating-system image: applications rely on identity, profiles, network services, licensing, endpoint devices, and sometimes a particular session model. A cloud desktop may change where those components run, how users connect, and how sessions share a host.
As an Amazon Associate I earn from qualifying purchases.
For Azure Virtual Desktop (AVD), Microsoft’s Cloud Adoption Framework migration guidance says there is no direct migration path from other VDI platforms because AVD has a different control plane. A deployment requires new host pools and session hosts, as well as decisions about images, profiles, and application deployment. Treat an existing image as useful input to an assessment, not as a complete migration plan.
Recommended Free Tools
The assessment process below applies broadly, but platform entitlements, supported operating systems, session capabilities, and configuration details are product- and time-specific. Verify them against current vendor documentation before committing to a design.
#1 Best Overall
What should the assessment establish first?
Set scope and success criteria
Agree which user groups, applications, sites, and business processes are in scope. Name an application owner and support contact for every important workload. Record maintenance windows, busy periods, data sensitivity, recovery needs, and what acceptable performance means to users.
Mark applications already scheduled for replacement, no longer supported, or unsuitable for a cloud desktop. Assessment is not a commitment to migrate every item in the current estate.
Build a validated inventory
Start with an authorized software scan or an existing configuration management database (CMDB), then confirm the findings with application owners. Automated discovery can miss infrequently used software, undocumented dependencies, and workflows that only users know about. Microsoft’s AVD assessment guidance describes using an independent software vendor (ISV) partner scan to collect application data and help map applications to user personas; treat scan output as an input to validate, not as proof of compatibility.
Capture enough detail to answer these questions for each application:
- Ownership and use: What are the name, version, vendor, support status, business owner, user groups, criticality, and usage frequency?
- Installation and maintenance: Which executables, services, scheduled tasks, drivers, middleware, runtimes, registry settings, configuration files, and installation switches are required? How are patches, upgrades, and rollback handled?
- Operating system and sessions: Which Windows or other operating-system versions and 32-bit or 64-bit components are supported? Does the app support multi-session use, or does it assume one user per machine?
- Identity, licensing, and privilege: Does it depend on domain identity, machine identity, elevation, a license server, or a particular activation method? How are concurrent users licensed?
- Data and profiles: Where are user settings, caches, and working files stored? Does the app expect persistent local paths, a persistent desktop, or a particular profile behavior?
- Connections: Which databases, file shares, APIs, identity services, license servers, print services, and external endpoints does it contact?
- Devices and media: Does the workflow depend on printers, scanners, smart cards, USB or serial/COM devices, graphics, audio, video, or other client redirection?
- Network behavior: What are typical and peak traffic patterns, authentication round trips, and known latency or bandwidth constraints? Where are the application’s back-end services located?
How should users be grouped before sizing cloud desktop pools?
Group users by comparable applications and technical needs rather than sizing from an organization-wide average. A persona should describe a real work pattern, such as a call-center user with a shared application set or an engineer who needs graphics acceleration and a dedicated desktop.
For each persona, record the factors that can change the pool design:
- Whether users need pooled or dedicated desktops, and the reason: persistence, security, compliance, performance isolation, or another constraint.
- Application mix, operating-system and session-model needs, profile behavior, data location, and peripherals.
- CPU, memory, storage, and graphics demands, using measured load where available rather than guesses.
- User count, expected peak concurrent sessions, geography, working hours, endpoint types, and support expectations.
Pool type, density, performance, graphics needs, Azure region, user count, and peak sessions all affect host-pool configuration. Microsoft’s older assessment article uses illustrative pool and density examples; those examples are not sizing recommendations for your environment. Measure representative workloads and validate the design in a proof of concept.
How do you check application and session compatibility?
Build a compatibility matrix with one row per application and a column for each candidate target environment. Check the application vendor’s current support statement against the exact target operating system, image, and session model. For AVD, explicitly verify whether the app is supported on Windows Enterprise multi-session if that is the planned host configuration. A successful launch alone does not establish compatibility.
Test the complete business workflow, including concurrent use where relevant, data integrity, authentication, installation and updates, profile persistence, printing and devices, and recovery from a failed session or host. Check drivers and services especially carefully: they may require machine-level access or behavior that conflicts with a shared multi-session host.
If an application does not work in the intended multi-session configuration, compare viable alternatives rather than forcing it into the pool:
Rank #3
- Use a dedicated or personal pool if the application’s support, isolation, or persistence needs justify it.
- Remediate the image, installer, configuration, or supporting components, then retest.
- Upgrade or replace the application if the vendor supports a suitable current version.
- Retain the workload on-premises or in a hybrid placement if its dependencies make cloud hosting unsuitable.
- Retire it if the business owner confirms it is no longer needed.
How do you test latency and the full network path?
Test both the user-to-desktop connection and the desktop-to-application dependencies. They are separate paths: a responsive desktop connection does not mean a legacy application will perform well if its database or file service is far away.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
For AVD, Microsoft’s prerequisites recommend client-to-region round-trip time (RTT) below 150 ms to the Azure region that contains the host pool and recommend locating session hosts near users. This is Microsoft’s platform guidance, not a guarantee that every application will meet its own response-time needs below that threshold. Separately measure session-host paths to databases, file services, identity and directory systems, license servers, APIs, print services, and security inspection points.
Test representative offices, remote locations, endpoints, and busy periods. Record RTT, packet loss, throughput, and jitter when interactive media is involved, alongside application response time during actual workflows. If a critical back end is remote, evaluate whether to keep it near users, move it with the desktop workload, optimize the connection, or retain a hybrid design. Validate the choice with the application vendor and business owner.
What identity, profile, security, and operating details belong in scope?
Document the dependencies and responsibilities that determine whether the service can be operated safely and consistently:
- Domain and identity requirements, authentication flows, privileged access, and endpoint posture.
- Profile storage, expected profile size, persistence, data residency, and recovery behavior.
- Image ownership, application deployment, patching, monitoring, backup, and recovery procedures.
- Security controls, data handling, support boundaries, and responsibility for incidents.
- Where user data and application back ends will live, and which network paths must be protected.
For an AVD migration, Microsoft’s migration guidance discusses either migrating existing profiles—for example, to Azure Files for FSLogix containers—or creating new profiles, alongside deploying applications and building new host pools. That is an example approach, not a requirement for every cloud desktop platform or environment.
Free tools Windows power users keep installed
One-click scans. No signup required.
How should licensing and vendor support be checked?
For the desktop operating system, each application, database, middleware component, and VDI platform, identify the license owner, assignment model, virtualization and cloud rights, support lifecycle, and any external-user terms that apply. Do not infer cloud or multi-session rights from an existing on-premises agreement. Get written confirmation from the relevant vendor or a licensing specialist when terms are unclear.
Microsoft’s AVD planning guidance describes access through eligible existing licenses or external-user access pricing, while virtual machines incur base compute charges. Eligibility depends on current Microsoft terms and the organization’s agreements; verify the applicable terms and cost model for the planned deployment rather than treating a general description as a quote or entitlement determination.
Also verify endpoint, peripheral, and collaboration requirements against the exact client and host configuration. These details change. For example, Microsoft’s Teams VDI documentation, accessed in 2026, says WebRTC-based optimization for Windows endpoints connecting to Citrix and AVD/Windows 365 reached End of Support on October 1, 2026, and is scheduled for End of Availability on April 1, 2027. Microsoft says that after end of availability the optimization stops working and the new optimization is enforced; if optimization fails, Teams falls back to server-side rendering. The same documentation lists Windows Server 2016 and 2019 as unsupported for the Teams desktop client. Confirm current milestones and requirements before designing or releasing a Teams-dependent workload.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What should a representative proof of concept test?
Microsoft’s AVD planning guidance calls for a proof of concept before first deployment. Its assessment guidance also cautions that a baseline proof of concept is unlikely to be production-ready as-is. Use the exercise to test assumptions and expose work, not to declare the entire estate ready based on one successful login.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsChoose applications that represent both business importance and technical risk. Include different personas, locations, endpoints, peripherals, and concurrency levels that approximate busy use. Define pass/fail criteria before testing, with application owners involved in deciding what acceptable performance and recovery mean.
Best Value
- Can users complete the agreed business workflows and preserve data correctly?
- Does the application run on a supported OS and session configuration, including expected concurrent use?
- Do profiles, settings, authentication, updates, and licensing behave as expected?
- Do required printers, smart cards, USB devices, audio/video, and other peripherals work from intended endpoints?
- Are user-to-host and host-to-back-end response times acceptable at representative locations and busy periods?
- Do security, monitoring, support, backup, and recovery procedures work as designed?
- Does the operating and cost model fit the organization’s requirements?
Run the proof of concept in the intended cloud region and with the endpoint types users will actually use. Log defects, workarounds, owners, and retest results so that decisions are based on evidence from the planned configuration.
How should each application be classified after testing?
Record one decision per application, with an accountable owner, required actions, dependencies, and a target sequence. Use these outcomes to turn findings into a migration plan:
| Decision | Use it when | Next action |
|---|---|---|
| Migrate as-is | The supported target configuration passes workflow, performance, dependency, and operations checks. | Schedule it with the appropriate persona and pilot cohort. |
| Remediate, then migrate | A fix to the image, installer, configuration, profile, or supporting component is feasible. | Assign the fix and require a successful retest before rollout. |
| Upgrade or replace | The current version is unsupported or cannot meet target requirements, but a suitable alternative exists. | Plan the application change with its owner and users. |
| Isolate in a dedicated pool | The application needs a personal desktop, specific OS, isolation, or session behavior that a shared pool cannot provide. | Validate the dedicated design, its support, and its operating implications. |
| Retain on-premises or hybrid | A critical dependency, support condition, or measured experience makes full cloud placement unsuitable. | Document the ongoing connection, ownership, and review conditions. |
| Retire | The business owner confirms the application or workflow is no longer required. | Follow the organization’s data, access, and decommissioning controls. |
Prioritize rollout by business value, technical risk, dependency readiness, and user impact. Start with pilot cohorts, monitor real experience, and use what happens in operation to update pool sizing and subsequent release plans.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →How should cloud desktop options be compared?
When evaluating more than one target, compare the same requirements against each option rather than relying on a broad feature claim. Useful axes include:
- Application fit: supported operating systems and session models, legacy dependencies, packaging, updates, and vendor support.
- User experience: measured user and back-end latency, graphics, audio/video, peripherals, and endpoint coverage.
- Architecture: pooled versus personal desktops, profiles, identity, data placement, region, and hybrid needs.
- Operations: image lifecycle, monitoring, patching, recovery, support model, and required administrator skills.
- Commercial fit: entitlement, compute, storage and network consumption, management, and remediation costs.
Microsoft’s AVD planning material presents AVD and Citrix DaaS as options and describes Citrix DaaS as supporting a broader range of identity providers, operating-system types, peripherals, collaboration solutions, and locations, including existing datacenters. Treat this as a vendor/platform description to validate against your actual requirements, not as a universal recommendation.
Which Microsoft guidance should be checked for current requirements?
The assessment process is durable, but implementation details and support terms can change. Microsoft’s Cloud Adoption Framework article “Assess Azure Virtual Desktop,” updated August 27, 2026, is marked deprecated, is no longer being updated, and says it will be removed on October 30, 2026. Use current product documentation and architectures for implementation decisions, and verify current operating-system, endpoint, collaboration, and licensing requirements before deployment.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
