Zero‑Downtime Migration from a Hosted Data Center to Azure Virtual Desktop
A real‑world technical case study detailing how a four‑server VMware environment was retired and replaced with Azure Virtual Desktop (AVD) and Azure Files—built from an empty tenant to users working in production in ~14 days, with no business interruption.
What You’ll Learn
This white paper walks through the migration sequence end‑to‑end—from first tenant setup to the final shutdown of legacy servers—plus the issues you only find in real deployments.
Subscription/RBAC, logging, virtual network, and security baselines established before workloads are introduced.
Microsoft Entra ID users/groups, MFA, conditional access, and access patterns designed without relying on an on‑prem domain controller.
Two session hosts split across zones, outbound‑only connectivity, and validation of brokering and session distribution.
Container format verification, attach performance tuning, and proving roaming across both hosts before app rollout.
Separate file shares for profiles vs. business data, authenticated with Entra Kerberos for cloud‑only identities.
Examples include Defender attach scanning stalls, legacy drive‑letter conflicts, VPN “false success”, and storage billing decisions.
Project Timeline
A practical two‑week pattern: the environment is built in parallel with pilot usage. The first user signs in around day 7 while the rest is still being assembled.
Discovery + foundation: scope, acceptance criteria, subscription/governance, baseline monitoring.
Identity + network: Entra groups, MFA/CA, vNet/subnets, egress and DNS decisions.
Storage + AVD build: Azure Files (profiles + data), Entra Kerberos, host pool and session host provisioning.
First pilot user: validate sign‑in, profile attach, desktop launch, printing, and core workflows.
Hardening + apps: FSLogix tuning, Defender exclusions validation, app packaging, edge case remediation.
Staged cutover + handover: migrate users in batches, verify data/apps, document runbooks, decommission plan.
The Results
Proven Outcomes
The approach prioritises parallel build + staged cutover so production remains available throughout. Legacy servers are only retired after acceptance criteria are met.
Unplanned downtime during the migration window
Legacy server roles replaced (after verification)
Standardised desktop delivery with FSLogix profiles
Runbooks + architecture documentation delivered