Customers may experience slow performance, session delays, or profile load issues when using certain Azure VM sizes (e.g., Standard_D16s_v3 and Standard_D16as_v4) in Windows 11 Enterprise multi-session deployments.

When attempting to launch four or more concurrent published applications on a Windows 11 multi-session host (Azure), sessions may fail to start correctly.

The affected VM sizes show limitations in disk I/O performance, particularly in write IOPS and throughput.

For example:

• • Standard_D16s_v3: Write IOPS is limited (~187), making it unsuitable for profile write-heavy workloads.

  • Standard_D16as_v4: Write performance is improved (~32,000 IOPS / 800 MB/s), but read throughput is lower (~255 MB/s), which can still become a bottleneck.

These limitations result in performance degradation when user profiles or applications require higher disk I/O.

Observed Results

Non-working / performance issues:

• • Standard_D16s_v3

  • Standard_D16as_v4

Working / stable performance:

• • Standard_E8bds_v5 – offers significantly higher read/write throughput (up to ~60,000 IOPS / 1200 MB/s), ensuring stable performance in multi-session workloads.

VM Size Selection

• Avoid v3 for Windows 11 multi-session workloads with Citrix.

• Prefer VM families with higher I/O capabilities, such as v5 series, which are Generation 2 VMs.

• Generation 2 VMs support Paravirtualization (PV) drivers, enabling more efficient virtualization and significantly improved I/O performance compared to Generation 1 VMs.

Reference Citrix Design Guidance

Citrix does not provide a single “minimum specification” that guarantees performance across all scenarios.

Instead, refer to:

• Citrix Tech Zone – Single Server Scalability

This document explains how workload type, user profiles, and applications affect scalability, and should be used as a guideline during PoC and performance testing.

Reference Microsoft Documentation

Microsoft provides VM recommendations for multi-session workloads:

• Remote Desktop Services VM recommendations

Monitor and Validate I/O Performance

• Use Azure Monitor for VMs to track read/write IOPS, throughput, and latency.

• Validate chosen VM sizes through PoC and load testing in your specific environment.

IOPS / Throughput Guidance

• There is no single minimum MB/s or IOPS value that can guarantee acceptable performance across all workloads—workload profiles, concurrency, and I/O patterns all differ.

• We recommend using v5 series VMs, which generally offer a superior value proposition for I/O-intensive workloads, providing higher throughput, better scalability, and support for newer features (such as Gen2 VMs with PV drivers).

• Customers should simulate their intended deployment using stress testing tools and real-world usage scenarios.

  • Validate not only under steady state, but also during login/logoff peaks.

  • Vary load sizes (number of users, app usage, I/O load) to reveal weak points.

  • Confirm that the system is responsive, resilient, and meets user expectations under those varied loads.

  • The number of users successfully completing the multi-session test provides a key performance indicator under real-world conditions. This value - referred to as the VSImax session count - is used for the comparative analysis. The Login VSI workloads calculate the VSImax session count by observing the response time of a single user on the system. Please check the following article for further information:
    
    https://community.citrix.com/tech-zone/design/design-decisions/azure-instance-scalability/

Workaround

While evaluating alternative VM sizes, the following Citrix policy adjustments may help improve session stability (requires testing in customer environment for optimal values):

• Concurrent logon tolerance → Increase from default

• Logoff checker startup delay → Extend beyond default

These settings may reduce stress during peak concurrent logon/logoff operations.

Customers may experience logon failure due to performance issues.