AVD Assessment and Management User Guide

Overview

The AVD Assessment and Management DEX Pack provides visibility into resource usage, sizing efficiency, and operational issues across your Azure Virtual Desktop (AVD) environment. Use this dashboard to monitor performance, identify inefficient resource allocation, and prioritize remediation actions using data from real user sessions.

The dashboard is organized into three main areas:

Sizing Recommendation: Determines whether Azure Virtual Machines are over-provisioned, under-provisioned, or right-sized based on resource utilization.
Session Host Performance: Provides visibility into session host performance, including resource utilization trends, per-host and per-user averages, and detailed session-level metrics.
Error Summary: Shows insight into VM session errors and identifies machines and users needing quick fixes.

How This Feature Helps You

This DEX Pack helps IT and EUC teams:

Optimize AVD host sizing to balance performance and cost
Identify hosts that need additional capacity before users are impacted
Find over-provisioned systems that may be candidates for cost reduction
Understand which users and applications drive high resource usage
Investigate session errors and connection failures quickly

Together, these insights help you improve user experience while controlling infrastructure spend.

Sizing Recommendation

Use the Sizing Recommendation dashboard to evaluate CPU and memory allocation for AVD session hosts:

Set the CPU and memory sizing options at the top of the page.
Review the colored breakdown chart.
Select a section of the chart to filter results.
Review the system list to see which hosts fall into the selected category.
Review the suggested CPU and memory changes for each host.
For systems flagged as insufficient, review the application list to understand which workloads contribute to resource pressure.

Recommendation Logic

The sizing engine evaluates CPU and memory independently and then produces an overall recommendation.

CPU recommendations

Increase CPU: The current core count is below the target range.
Decrease CPU: The current core count exceeds the target range.

Memory recommendations

Increase Memory: The current memory allocation is below the target range.
Decrease Memory: The current memory allocation exceeds the target range.

Overall recommendation logic

If either CPU or memory requires an increase, Overall = Increase Resource.
If neither requires an increase but at least one requires a decrease, Overall = Decrease Resource.
If neither CPU nor memory requires a change, Overall = No Change.

Suggested ranges

CPU suggested range: From the minimum number of cores required to stay below the upper utilization target, to the maximum number of cores that still stay above the lower target.
Memory suggested range: From the minimum memory required to stay below the upper utilization target, to the maximum memory that still stays above the lower target.

Default Thresholds

The default CPU and memory sizing thresholds are defined in Hardware Refresh.

CPU Sizing

CPU sizing methods:

Base sizing on average active utilization — when a user is logged in and the screen is unlocked
Base sizing on maximum utilization
Base sizing on average utilization

The CPU Lower (%) and CPU Upper (%) thresholds control how strict the sizing classifications are that categorize CPU as Increase Resource, Decrease Resource, or No Change.

Memory Sizing

Memory sizing methods:

Base sizing on average active utilization — when a user is logged in and the screen is unlocked
Base sizing on maximum utilization
Base sizing on average utilization

The Memory Lower (%) and Memory Upper (%) thresholds control how strict the memory sizing classifications are that categorize memory as Increase Resource, Decrease Resource, or No Change.

System Category Breakdown by System Resource

This chart shows the distribution of systems by sizing recommendation across CPU, Memory, and Overall.

Each category is divided into color-coded states:

Green – No Change. Systems appear appropriately sized.
Orange – Increase Resource. Systems may need additional capacity.
Blue – Decrease Resource. Systems may be over-provisioned and candidates for reduction.

Use the chart as a filter, not just a summary:

Start with Overall.
Identify the largest orange segment.
Select that segment to filter the system list.
Review the hosts driving the result.
Repeat for blue if cost optimization is a goal.
If needed, switch to CPU or Memory to focus on one resource at a time.

System List and Details

The color-coded system list shows detailed sizing and utilization data per session host and supports filtering from the breakdown chart:

Recommendation colors (Green, Orange, and Blue): Indicate the suggested sizing action.
Status colors:
- Green – Online systems
- Yellow – Cached data
- Red – Offline systems

Displayed fields include:

System name
Status
Operating system
Users in last 30 days
Memory capacity (GB)
CPU cores
Overall recommendation
CPU recommendation and suggested range
Memory recommendation and suggested range
Max, average, and average active CPU usage
Max, average, and average active memory usage

Application Resource Consumption

This section shows resource usage for user-launched applications. Use this view to identify applications that contribute to CPU or memory pressure.

You can filter by system or chart selection to review:

Application name
Number of users
Average CPU usage (MIPS and cores)
Average memory usage (MB)
Average disk I/O reads and writes
Average network usage (Bytes/s)
Total application starts
First and last use timestamps

Session Host Performance

Use this dashboard to analyze performance trends and user impact across session hosts:

Filter the dashboard by host group.
Review health impact trends.
Review CPU and memory trends per host.
Analyze per-host resource utilization.
Select a host to inspect its user sessions.
Select a user to review sessions across hosts.

Host CPU & Memory Usage Trending

Shows average daily CPU and memory utilization per host over the last 30 days:

Represents total host usage
Not session-level data

Shows average daily health impact per host over the last 30 days:

Represents total host health impact
Not session-level data

Use the Metric filter to switch the chart between Health Score, Host Count, and impacted-minute views for CPU, memory, disk, network, latency, startup, virtual memory, virtualization, installations, updates, events, faults, and hardware issues.

Resource Utilization per Host

Lists session hosts and shows average per-session resource usage over the last 30 days:

Host
Average CPU usage per session
Average memory usage per session
Average disk usage per session
User count
Session count

Selected Host Sessions

Shows all user sessions on the selected host from the last 30 days, including:

User name
Session start and end
Session duration
Client name and IP
Average CPU, memory, and disk usage
Session timing metrics such as initialization and readiness

Resource Utilization per User

Shows average resource usage per user across all sessions:

User name
Average CPU usage
Average memory usage
Average disk usage

Selected User Sessions

Shows all sessions for the selected user, including:

Host
Session timing and duration
Client details
Average resource usage
Session initialization and readiness metrics

Error Summary

Error Summary - Click for Details

This pie chart shows the distribution of AVD session-related errors for the selected time range.

You can:

Select Internal or External error types
Adjust the date range

Internal vs External Errors

This chart shows the volume of internal and external errors over time.

Select a data point to investigate errors that occurred at a specific time.

Details for Selected Errors

The details table lists information for the selected errors, including:

User principal name
Error time
Activity type
Client details
Session host name and IP address
Connection type
Session host OS version
Session host agent version
Gateway region

Use this view to diagnose connection issues and identify patterns affecting specific users or hosts.