Azure Virtual Desktop Host Pool Auto-Scaling

1. Scaling Fundamentals & Session Host Caps

Before enabling autoscale, set the session host caps that define how aggressively the platform can compress the pool. These values drive both power management and dynamic autoscaling:

• MaxSessionLimit: Configure this per host pool - the default is unsuitable for autoscale and must be replaced with a deliberate concurrency target.
• Minimum / Maximum Hosts: Define the floor that keeps the fleet responsive and the ceiling that contains spend.
• RBAC: Grant Azure Virtual Desktop power-state management rights so session hosts can be started and stopped automatically.

2. Time-Based Autoscale Scaling Plans

Scaling plans apply time-based schedules that match host availability to business hours. Each plan holds one or more schedules per day with peak and off-peak windows, and every assigned host pool must reside in the same Azure region as the plan.

3. Usage-Based Dynamic Autoscale

Dynamic autoscale goes beyond the clock: it watches real session demand and adjusts the running host count continuously. It applies to pooled host pools with session host configuration and pairs naturally with fluctuating shift patterns.

• Demand Tracking: Session buffer and drain thresholds react to live user counts instead of a fixed schedule.
• Crossover Sessions: Account for shift overlap by analyzing historical concurrency before tuning the buffer.
• Availability: Confirmed in Azure; check feature availability for your region and Government environments before adoption.

4. Custom Scaling with Azure Automation

For environments that predate autoscale or need bespoke logic, the Azure Automation scaling runbook plus a Logic App automates start/stop of session hosts. Remember the guardrail: autoscale and the scaling script cannot run on the same host pool.

• Runbook Deployment: Deploy the runbook with Contributor rights and configure managed identity permissions for power management.
• Logging: Optionally wire a Log Analytics workspace and query WVDTenantScale_CL for host and session counts.
• Maintenance: Keep the runbook version current and validate the day after Windows Azure maintenance windows.

5. Verification, Testing & Cost Analysis

Validate the plan in a pilot pool before broad rollout: watch disconnects, logon latency at pool minimum, and cost per session host across a full billing cycle.

• Synthetic Load: Drive a scripted login wave against the pilot pool and confirm hosts spin up inside the schedule.
• Insights Dashboards: Use Azure Virtual Desktop Insights to review start/stop events and session host drain behavior.
• Iterate: Adjust buffer sizes weekly; log cost per host and user-satisfaction signals together to avoid over-optimizing.