The Autoscale Engine monitors session load across your host pools and automatically provisions or deallocates session hosts to keep capacity matched to demand. Scaling can be driven by real-time load, by schedule, or by both together, so pools scale out ahead of the morning logon wave and scale back in once users log off. No scripting is required; scaling behavior is configured per host pool and applied continuously.
Key capabilities
- Load-based scale-out that adds hosts before session demand saturates capacity
- Schedule-based scaling windows for predictable daily and weekly usage patterns
- Configurable per-host-pool thresholds, host minimums, and cooldown windows
- Automatic scale-in that deallocates idle hosts once load drops
- Every scale-out and scale-in decision logged with before/after host count for audit
- Works alongside Auto On/Off Scheduling and Idle Timeout Shutdown for layered cost control
How Autoscale Engine works
- Continuously polls session count, capacity utilization, and drain state across each host pool
- Compares live load against the pool's configured thresholds and schedule windows
- Provisions additional hosts when demand rises and deallocates them once it falls
- Respects cooldown windows to avoid thrash and logs each decision to the operations timeline
Available on: Atlas AVD plan and above
Autoscale Engine is included in every Atlas AVD subscription.