Table of Contents

• Support Windows 10 and requirements for Windows 11
• Setup and configuration
• Hypothesis and results
• Conclusion

Windows 10 is rapidly reaching its end-of-life (EOL) date in October 2025. That means that organizations that are still running on Windows 10 should start preparing to migrate to Windows 11 quickly. This transition is particularly significant for Virtual Desktop Infrastructure (VDI) environments, where minor differences in resource consumption, user experience, and system performance can substantially impact scalability and costs. What should organizations consider when planning a large-scale migration for their VDI workloads from Windows 10 to Windows 11?

Support Windows 10 and requirements for Windows 11

Windows 10 will reach its support milestone with the end of support on October 14, 2025. After this date, Microsoft will no longer provide mainstream security updates, bug fixes, or feature enhancements for any edition of Windows 10: Home, Pro, Enterprise, and Education. The final feature update for the operating system is version 22H2, the last supported release. For those who can make the date, there is an option to extend support, but that comes with an additional cost.

The Long-Term Servicing Branch (LTSB) and Long-Term Servicing Channel (LTSC) editions of Windows 10 are designed for specialized devices that require a more stable operating system without frequent feature updates. These two editions have an extended support lifecycle compared to the other Windows 10 versions:

• Windows 10 Enterprise 2015 LTSB: Extended support ends on October 14, 2025. ​
• Windows 10 Enterprise 2016 LTSB: Extended support continues until October 13, 2026.

Microsoft offers Extended Security Updates (ESU) for organizations that do not want or cannot upgrade to Windows 11. Upgrading to Windows 11 requires meeting specific hardware requirements, which some older PCs may not fulfill.

Windows 11 has extended requirements over Windows 10:

Upgrading Windows 11 is not just a recommendation; it is necessary to remain compliant and continue to qualify for support. Not upgrading to Windows 10 can leave systems vulnerable to security threats. Without ongoing updates, vulnerabilities could be exploited, increasing the risk of cyberattacks and data breaches.

Setup and configuration

This research has taken place in the GO-EUC lab environment. To return the impact of both Windows 10 and Windows 11, the following scenarios are included:

• Windows 10 22H2
• Windows 11 24H2

The scenarios are served using Citrix Virtual Apps & Desktops, version 2402 LTSR, supported on Windows 10 and Windows 11. More information about CVAD 2402 LTSR can be found here.

Each virtual machine is equipped with 4vCPUs and 4GB of memory. Please note that Microsoft states that 4GB is the minimum required memory to run Windows 11.

A pool of 32 machines is hosted, ensuring the host will not reach peak utilization and, therefore, inadvertently create unwanted bottlenecks. Both images were patched with the latest available patches and optimized using the latest Citrix Optimizer 3501; this release of the Citrix Optimizer includes templates for both used operating systems.

Please note that Microsoft Teams is installed on Windows 11 by default. The default start-up is disabled to ensure it can be compared to Windows 10.

All simulations are executed with LoadGen using the default GO-EUC workload. The methodology can be read here for more information about the testing used.

Hypothesis and results

Based on previous comparisons between the performance and scalability of Windows 10 and Windows 11, the initial hypothesis is that there should only be a minimal impact on performance and scalability between Windows 11 and Windows 10. That being said, Windows 11 introduced additional background services (such as enhanced telemetry, cloud integrations, etc.) that could, in theory, have extra overhead. Furthermore, Windows 11 has more visual effects than Windows 10, including transparency effects and dynamic animations; however, due to the optimizations done before the tests, these will have less impact.

This research took place in the GO-EUC lab, so the results included performance data at both the Hypervisor and VM levels. Let’s start with the hypervisor.

Overall, there is a slight increase in CPU usage observed, with Windows 11 reporting a small increase in overall CPU use on a hypervisor level.

This increase in overall CPU usage may affect host scalability in multi-session environments, and this could reduce the number of concurrent users per VM.

Memory usage is not a differentiator, as the allocated memory on the hypervisor is similar. This is generally expected as both deployments have been assigned 4GB of memory per VM. This means that VM sizing for RAM can remain the same between the two OS versions.

In both reads and writes, Windows 11 has higher spikes compared to Windows 10, but on average, it’s very even between both scenarios. As memory consumption isn’t a differentiating factor based on memory metrics, it can assumed that the increased IOPS is not related to memory swapping or paging. Windows 11 could simply be writing more data to the event logs; however, this was not investigated further.

If IOPS capacity is a limiting factor, the increase in IOPS needs to be considered when scaling storage in preparation for migrations to Windows 11. IOPS performance issues can be very disruptive, where the entire session could be unresponsive, which can be detrimental to the overall user experience.

There is a significant difference in the network activity, with more transmitted data in the Windows 11 scenario. Windows 11 sends around 12% more data on average than Windows 10 (4,089 KB/s vs. 3,639 KB/s). The maximum transmitted data rate is also higher for Windows 11, suggesting a more active upload behavior. This could indicate that Windows 11 by default sends more telemetry or background data than Windows 10.

Note that a proxy is configured to prevent internet access on the VM level, but nevertheless, there is more activity. At this point, it is not clear what the root cause is, and it requires a deeper investigation.

Now, let’s switch to the VM perspective.

The difference noticed on the hypervisor level is also observed on the VM level, where Windows 11 requires more CPU usage than Windows 10. Depending on the CPU capacity and clockspeeds, your users might experience this in general use, but when your environment is sized accordingly, there should be no noticeable difference.

Windows 11’s overall memory consumption is higher compared to Windows 10’s. While the memory starting point is almost the same at the start of the test, it is structurally higher for the remainder of the test. On average, Windows 11 consumes 13% more memory than Windows 10.

Interestingly, on a VM disk level, Windows 10 has a slightly higher read footprint. There are clear spikes that affect the average comparison between both scenarios. Compared to the hypervisor-level metrics, these spikes are more visible on the VM metrics themselves. Because the impact on the hypervisor level is only minute, this will most likely not impact the scalability or performance.

The same pattern from a network-specific perspective is also visible on the VM-level metrics, where Windows 11 shows increased network activity compared to Windows 10.

Conclusion

Windows 10 is rapidly approaching its end of life, with official support ending in October 2025.

Customers who have not migrated to Windows 11 for the VDI infrastructure yet, or customers using Windows 10 on their endpoints and fat clients, need organizations to act now to ensure a smooth and successful transition to Windows 11. The shift is significant for Virtual Desktop Infrastructure (VDI) environments, where changes in resource consumption can directly impact performance, scalability, and overall user experience.

The data from this research concludes that while Windows 11 introduces only marginal increases in CPU usage, memory consumption, and network activity compared to Windows 10, these differences can still impact system density, application responsiveness, and cost efficiency. If your current hardware that hosts the Windows 10 VDIs is resource-constrained, consider evaluating these factors to avoid potential performance bottlenecks. In this research, we’ve seen that there is an increasment of 17% in CPU load and 13% in memory load. This is a good indicator that additional hardware might be required for your hypervisor platform when migrating.

Please note that Windows 11 out-of-the-box installations include Microsoft Teams. For this research, Microsoft Teams’ autostart was disabled. In real-world scenarios where Microsoft Teams is needed, this could have an additional impact. Customers should be ready to optimize the Teams deployment to reduce the impact of running Teams inside the VDI. See the Microsoft Teams documentation for more information.

Are you in the process of migrating to Windows 11? Ensure you validate your VDI infrastructure for Windows 11 readiness. Assess system resource constraints and plan for necessary optimizations to ensure a smooth transition. Make sure to focus on optimizing the deployment of Microsoft Teams within the VDI environment to minimize its impact on performance.

Photo by Jorge Tung on Unsplash