Examine host pools, session hosts, and scaling

Completed

A virtual desktop design often fails for one of two reasons: too little capacity during peak hours or too much capacity during quiet hours. Host pools, session hosts and scaling features give you the controls to balance user experience and cost. This unit explains how the building blocks relate and how common scaling strategies behave in real environments.

Decision you make	Options you evaluate	What the decision controls
Host pool type	Pooled or Personal	Whether users share session hosts or receive dedicated desktops
Session distribution	Breadth-first or Depth-first (pooled only)	How new sessions spread across available session hosts
Scaling approach	Autoscale scaling plans, Start VM on Connect or custom automation	How many session hosts run at different times of day

Choose a host pool type that matches the user model

A host pool is a collection of Azure virtual machines registered as session hosts. The host pool type defines the user-to-VM relationship and drives decisions for profiles, scaling and cost.

Pooled host pools: shared capacity

A pooled host pool places many users on a shared set of session hosts. This model fits standardized application sets, task workers and high concurrency scenarios. Because users can land on different session hosts, this approach depends on a roaming profile solution.

Personal host pools: dedicated desktops

A personal host pool maps each user to a specific session host. This model fits users who need a persistent environment, predictable performance or applications that don't work in multi-session scenarios.

Host pool type	Session mapping	Typical benefits	Typical trade-offs
Pooled	Users connect to any available session host	High density, shared capacity, flexible scaling	Requires strong profile strategy
Personal	Each user connects to one assigned session host	Predictable performance, persistent desktop	Higher per-user cost

Understand session hosts and how you manage them

A session host is the compute layer that runs user sessions. How you manage session hosts determines how you patch, update and scale capacity.

Session hosts as a standardized fleet

A standardized session host fleet starts with a consistent image. When all session hosts share the same base configuration, scaling and troubleshooting become predictable.

Host pool management approaches

Azure Virtual Desktop supports different management approaches for session hosts.

• Standard management: You control VM creation, updates and scaling using your own tools.
• Session host configuration: The service stores a desired configuration and integrates creation, updates and scaling for pooled host pools.

Management approach	Where configuration lives	How scaling and updates behave	Best fit
Standard	You manage VMs directly	Scaling and updates follow your operational model	Environments needing customization
Session host configuration	Azure Virtual Desktop service	Service-managed creation and updates with autoscale integration	Pooled host pools that benefit from standardization (pooled-only, preview)

Control user experience with load balancing and session limits

Load balancing controls how new sessions land on session hosts in pooled host pools.

Breadth-first vs. depth-first

• Breadth-first spreads sessions across hosts to reduce contention.
• Depth-first fills one host to a maximum session limit before using the next.

Algorithm	Session placement	Best use case	Trade-off
Breadth-first	Even distribution	Performance-sensitive workloads	Reduced scale-in efficiency
Depth-first	Concentrated sessions	Cost optimization	Requires accurate session limits

Drain mode for maintenance

Drain mode prevents new sessions from landing on a session host while allowing existing sessions to finish. This approach supports patching, scale-in and host rotation.

Apply scaling options that match demand patterns

Scaling defines how many session hosts run at any time and how quickly capacity responds to demand.

The following diagram shows how these three design decisions—host pool type, load balancing, and scaling—build on each other.

Autoscale scaling plans

Autoscale uses schedules to power session hosts on or off. One scaling plan can apply to multiple host pools, but each host pool uses only one plan.

Autoscale also supports personal host pools through power management. Idle personal session hosts can hibernate or deallocate based on schedules, reducing compute costs while preserving user state.

Start VM on Connect

Start VM on Connect powers on session hosts only when a user connects. This approach reduces idle cost but increases initial connection time.

Demand pattern	Scaling option	Reason
Predictable office hours	Autoscale scaling plan	Schedule matches daily usage
Spiky sign-ins	Autoscale scaling plan with tuned limits	Handles rapid demand changes
Personal desktop cost optimization	Autoscale with personal pool power management	Reduces idle compute cost
Low usage with occasional access	Start VM on Connect	Hosts remain powered off until needed
Strict custom requirements	Custom automation	Enforces specialized rules