AVD pool - performance issues

Hi Krishnakumar_M,

Performance issues in AVD environments can indeed be challenging, and it sounds like you're dealing with a resource bottleneck, possibly memory-related as you mentioned. Let's break down the potential solutions and considerations for upgrading your pool configuration.

Current Configuration and Issues: You currently have four D8sv3 session hosts, each with 8 vCPUs and 32 GB of RAM. For 50 users, this setup can be stretched, especially if each user has a moderate workload.

Analyzing the Workload:

1. CPU Usage: As per the best practices mentioned by other community members, you should consider using only 80% of the total available cores to avoid overloading.
   • 4 x D8sv3 = 4 x 8 vCPUs = 32 vCPUs
   • 80% of 32 vCPUs = 25.6 vCPUs for 50 users, which is roughly 2 users per vCPU.
2. Memory Usage: Each D8sv3 has 32 GB of RAM, so with 4 hosts, you have 128 GB of RAM total. For 50 users, if memory usage per user is high, this could be a bottleneck.

Upgrade Options:

1. Eight D4sv3 VMs:

   • Each D4sv3 has 4 vCPUs and 16 GB RAM.
   • 8 x 4 vCPUs = 32 vCPUs; 80% of 32 = 25.6 vCPUs, still 2 users per vCPU.
   • Total RAM: 8 x 16 GB = 128 GB (no change in total memory).

2. Eight E4sv3 VMs:

   • Each E4sv3 has 4 vCPUs and 32 GB RAM.
   • 8 x 4 vCPUs = 32 vCPUs; 80% of 32 = 25.6 vCPUs.
   • Total RAM: 8 x 32 GB = 256 GB (doubling the total memory).

3. Two E16sv3 VMs:

   • Each E16sv3 has 16 vCPUs and 128 GB RAM.
   • 2 x 16 vCPUs = 32 vCPUs; 80% of 32 = 25.6 vCPUs.
   • Total RAM: 2 x 128 GB = 256 GB (doubling the total memory).

Recommendation: Given that you suspect memory is the major issue, and to avoid CPU bottlenecks:

• Eight E4sv3 VMs seem to be the most balanced choice. They double your current memory capacity to 256 GB, which should alleviate memory-related performance issues. The CPU allocation remains adequate with the 80% rule applied, and you avoid the complexity of managing fewer larger VMs (like E16sv3).

Additional Considerations:

• Monitoring Usage: Ensure you continuously monitor CPU, memory, and disk usage. Tools like Azure Monitor can help identify the exact resource causing the bottleneck.
• Application Impact: Verify if there are any additional applications running or if specific user activities are causing high resource consumption.
• Autoscaling: Implement autoscaling rules to dynamically adjust resources based on load.

I hope this helps!