In late 2025, a Houston construction company moved its entire IT backbone from Amazon Web Services to Microsoft Azure without a moment of downtime — an eight-app, 16-site migration completed in just five days. The successful cutover, orchestrated by Katy-based managed service provider (MSP) Impress IT Solutions, underscores the real-world hurdles of shifting legacy Windows workloads and field connectivity to the cloud while keeping operations humming.

For Windows administrators and IT pros, the project is more than a vendor success story; it’s a practical case study in how to handle the messy dependency chains — Active Directory, SQL databases, file-based apps, hybrid storage — that often turn a cloud migration into a minefield.

The nuts and bolts of the migration

According to Impress IT Solutions’ July 13, 2026 press release, the five-week project ran from August 1 through September 4, 2025. The MSP built a brand-new Azure tenant in parallel with the live AWS environment, using Nerdio for Azure management to deploy domain controllers, virtual networking, and an Azure VPN Gateway. A site-to-site VPN tunnel between the two clouds allowed data synchronization while the team tested every application and connection.

During the final sprint — a five-day cutover between August 31 and September 4 — the production workloads flipped from AWS to Azure in early-morning maintenance windows. Zero data loss and no planned downtime were claimed, though these assertions come from the MSP and have not been independently audited.

The application lineup is a who’s who of construction IT: Oracle/SQL-backed Primavera P6 for project management, PlanSwift for takeoffs, ACT! CRM, Microsoft Access databases, Egnyte scanning workflows, and a print server. Cloud-native integrations for Procore, Vista Viewpoint, and GCPay also made the trip. After the move, six AWS servers were decommissioned.

Why this matters for Windows and Azure admins

The project’s technical skeleton — Active Directory domain controllers, Azure VPN Gateway, and a zero-trust network access (ZTNA) overlay — will look familiar to any Windows admin who’s lifted an on-premises or co-located estate into Azure. But the devil is in the details.

First, the parallel-build strategy isn’t just a best practice; it’s what makes zero-downtime claims remotely plausible. By standing up a full Azure landing zone alongside the live AWS environment, the team could validate user profiles, Group Policy, DNS, and authentication flows before a single end user touched the new system. That validation step is critical when you’re dealing with legacy line-of-business apps that won’t tolerate half-baked connectivity.

Second, the remote-site reconfiguration is a masterclass in modern edge networking. The construction firm had 16 job sites and field offices using Cradlepoint routers. One location in Deer Park relied on Starlink in passthrough mode. Rather than extend a fragile, traditional VPN to these locations, Impress replaced the legacy access with Appgate SDP, a software-defined perimeter product that enforces identity-aware, session-level access. For Windows admins, that’s a signal: cloud migrations are an ideal time to ditch legacy VPNs and move to zero-trust models that integrate with Azure AD Conditional Access.

The performance reality check: cloud storage isn’t a silver bullet

Perhaps the most instructive troubleshooting episode involved PlanSwift, a construction takeoff tool that works with large, active project files. When those files sat solely in cloud storage, performance tanked for end users. The fix? A hybrid design: a local NAS device on-site to house active project data, backed up to the cloud.

This isn’t a failure; it’s a reminder that latency and bandwidth matter, especially for applications built around rapid file access. “Cloud-hosted” doesn’t automatically equal “high-performance.” For Windows admins planning Azure migrations, the takeaway is to profile every application’s I/O patterns. If an app expects low-latency access to shared files, test it under realistic load before the cutover. You might need Azure File Sync, a local cache, or — as Impress did — a simple NAS with cloud backup.

Another near-miss came during AWS decommissioning: an undocumented mapped network share was discovered on one of the servers slated for shutdown. The team recovered the data in time, but it’s a stark warning. Inventory your file shares, printer mappings, and data paths before you pull the plug on legacy servers. A common PowerShell script to enumerate shares or a tool like Microsoft’s File Server Resource Manager can save you from a nasty surprise.

The remote site puzzle: connectivity without chaos

The MSP’s field site work highlights how cloud migrations must account for both high-tech and low-tech constraints. Construction sites operate on spotty 4G/5G connections, satellite internet, and hardened routers. Impress’s use of Cradlepoint routers — known for SD-WAN capabilities — suggests the team built in failover and traffic management. For Windows admins, integrating such edge devices with Azure’s networking stack often means configuring Azure VPN Client profiles or deploying Azure Virtual WAN, but the press release doesn’t go into that depth. The bigger lesson: don’t assume reliable bandwidth; design for offline or sync-friendly operations where possible.

A blueprint for your own zero-downtime cloud migration

What can you steal from this project? Here’s a practical checklist distilled from the Impress IT Solutions playbook:

1. Build in parallel, then pivot
Spin up your target Azure environment — domain controllers, virtual networks, VPN gateways — while the runway is still long. Validate everything with a site-to-site VPN before moving a single workload.

2. Map every dependency
Don’t just list servers. Document file shares, printer mappings, DNS records, legacy apps with hard-coded IPs, and SQL connection strings. A surprise like the undocumented share in this project could mean data loss.

3. Performance-test real workloads
Applications like PlanSwift or Access databases that expect low-latency file access can break in a pure cloud model. Run synthetic and user-acceptance tests early, and be ready to deploy local caching (Azure File Sync, a NAS, or even a small on-prem caching server).

4. Rethink remote access
If you’re still on traditional VPNs, a cloud migration is the moment to move to ZTNA. Products like Appgate SDP, Microsoft’s own Global Secure Access, or third-party solutions integrated with Azure AD can give you granular, identity-based control over who reaches what.

5. Plan the final cutover in short bursts
Five early-morning windows over five days may be more manageable than one marathon weekend. It lets you roll back a single application if something goes wrong without affecting the whole business.

6. Decommission with care
Before shutting down old servers, run a final discovery pass. A simple Get-SmbShare or net share can reveal shares that nobody documented but that someone still uses.

The bigger picture: why construction and field-intensive industries are moving to Azure

This migration isn’t an isolated event. Construction firms, engineering shops, and field-service companies increasingly look to Microsoft’s cloud because it tightly integrates with the tools they already use: Windows devices, Office 365, Teams, and line-of-business applications built on .NET and SQL Server. The Houston project mirrors a broader shift away from AWS’s laissez-faire IaaS model toward Azure’s “landing zone” approach, which can feel more coherent for organizations that live and breathe Active Directory and Group Policy.

Moreover, the inclusion of Cradlepoint, Starlink, and Appgate SDP signals that cloud migrations are now inherently edge-to-cloud plays. It’s not just about moving VMs; it’s about re-architecting how remote workers and job sites connect while maintaining security and performance.

What to watch next

The MSP promises a full technical white paper on its website, which may deliver more granular data on the Azure VPN Gateway configuration, Nerdio policies, and Appgate SDP settings. For Windows admins, that document could be a goldmine. In the meantime, if you’re considering an AWS-to-Azure — or any major cloud — migration, start by cataloging your application I/O demands and your remote access points. The Houston project’s success was built on meticulous pre-work, not magic.