Windows 365 Cloud PC Latency Issues

Question

Hello,

 

We’re currently using several Windows 365 Cloud PCs to access our Microsoft 365 GCC High tenant. The machines are configured with Windows 365 Enterprise (2 vCPU / 8 GB RAM / 256 GB storage, GCCH subscription per user).

 

We’ve been experiencing noticeable latency during routine tasks. For example, when saving a Word document, it takes approximately 17 seconds just to browse to the destination folder and another 17 seconds to complete the save, about 34 second lag time total.

 

The delay is even more pronounced when converting Word documents to PDF. It takes around 17 seconds to navigate to the folder, an additional 48 seconds to save the PDF, and roughly 1 minute and 6 seconds total before the file appears on screen.

These delays are significantly impacting usability, even for basic day-to-day operations.

 

While monitoring Task Manager and watching the cursor spin for 45 seconds, I do not see the CPU or Memory max out when saving documents, but the memory is almost maxed out with no apps or browsers open, and the CPU does max out at times.

 Is the latency an issue across all Cloud PCs regardless of the specs?

We want to provision five additional Cloud PCs If we can solve the latency issue. We are looking the lowest-cost option: Windows 365 Enterprise (2 vCPU / 4 GB RAM / 64 GB storage, GCCH subscription per user) but we question if the low-end spec cloud PC even usable or is it just going to freeze up?

Thank you

Answer 1

Hello.

First of all, please check your background processes, as they also consume memory, not just active applications. The issue may be caused by tasks running in the background.

When the CPU hits 99% during simple folder navigation, it seems struggling with real-time security scanning or background tasks that are significantly more resource-intensive in a GCCH environment. And for a stable professional experience, you should upgrade these users to a 4 vCPU / 16 GB RAM tier to provide the necessary ceiling for Microsoft 365 apps to function without constant resource contention.

If you find my answer useful, please mark it as the "Accepted Answer" so other members can find the solution easily. Your help in supporting the community is much appreciated!
Best regards,
Scott Nguyen

Answer 2

The behavior described isn’t expected for all Cloud PCs and isn’t inherent to Windows 365. Latency and “freezing” are typically caused by a combination of:

• Where the Cloud PC is provisioned vs. where users and application data reside
• Network routing and protocol usage
• Undersized Cloud PC resources for the workload

From the screenshots, CPU and memory are already heavily utilized on a 2 vCPU / 8 GB Cloud PC with only light usage. Moving to a lower spec (2 vCPU / 4 GB / 64 GB) would further constrain resources and is likely to worsen responsiveness, not improve it.

To improve performance and assess whether additional Cloud PCs will be usable, consider these areas:

1. Check Cloud PC region vs. user and data location
   • Provision Cloud PCs in regions that are geographically close to either the users or, preferably, the application data (for example, Microsoft 365 data location).
   • When user, Cloud PC, and app data are far apart, latency increases for file operations such as opening/saving Word documents or exporting to PDF.
   • If Cloud PCs were provisioned in a distant region, update the provisioning policy to move them closer to the application data. Note that changing the region removes existing restore points, so plan timing accordingly.
2. Optimize network connectivity and routing
   • Ensure Cloud PCs use local and direct internet egress so traffic reaches the nearest Windows 365 RDP gateways and TURN relays quickly, then traverses Microsoft’s backbone. Avoid hairpinning RDP traffic through centralized VPNs, proxies, or TLS inspection devices, as these add latency and can cause poor desktop performance and delayed user input.
   • Prefer UDP for RDP sessions. RDP over UDP (with URCP and RDP Shortpath) provides lower latency and better handling of packet loss than TCP. In customer-managed networks (Azure Network Connections), RDP Shortpath is enabled by default and establishes a direct UDP-based connection where possible.
   • Reduce packet loss and jitter on the path between endpoints and the Cloud PC; both directly affect responsiveness.
3. Use Windows 365 monitoring tools
   • Review the Connection Quality Report for affected Cloud PCs to see round-trip time (RTT), available bandwidth, and protocol details. High RTT or low bandwidth will correlate with the delays observed when saving or converting documents.
   • Use these metrics to validate whether changes to routing, bandwidth, or region placement improve performance.
4. Right-size Cloud PC resources
   • The current 2 vCPU / 8 GB configuration is already close to its limits based on Task Manager screenshots. For typical office workloads with multiple browser tabs and Office apps, 4 GB RAM is generally insufficient and will increase paging and CPU contention.
   • If the existing 2 vCPU / 8 GB Cloud PCs are struggling, the 2 vCPU / 4 GB / 64 GB option is likely to be less usable, especially when running Microsoft Edge plus Office apps and converting documents to PDF.
   • If performance issues persist after addressing region and network, consider resizing existing Cloud PCs to a larger SKU (more vCPU/RAM) rather than downsizing. Windows 365 Enterprise supports remote resizing of vCPU, RAM, and storage.
5. Ongoing optimization
   • Continuously monitor latency, bandwidth, and error rates, and adjust Cloud PC placement and network configuration as needed.
   • For users with more demanding workloads or who frequently work with large documents and PDFs, profile them as “power users” and assign higher-spec Cloud PCs and better network paths.

In summary, the latency is not a universal Cloud PC limitation. It is primarily influenced by region placement, network path quality, and resource sizing. For the planned additional Cloud PCs, the lowest-cost 2 vCPU / 4 GB / 64 GB configuration is unlikely to meet the needs described; focus first on optimizing region and connectivity, and then size Cloud PCs at or above the current 2 vCPU / 8 GB level for acceptable usability.

---

References:

• Optimal Provisioning of Cloud PCs: Considerations based on User Location, Application Data Location and Network Optimization
• Windows 365 Connectivity Overview
• Relative performance for different Cloud PC sizes