Most infrastructure failures are abstract to the people affected by them. Azure Virtual Desktop failures are not. When a session host degrades, someone loses the machine they were working on, mid-task, and the support ticket arrives before your monitoring does.

Install the Gremlin Agent on your session hosts and you can find these problems on a quiet afternoon rather than a Monday morning. With Gremlin, you can:

• Confirm sessions survive a host failure. Shutdown experiments remove a session host while users are connected. You'll learn whether sessions fail over to a healthy host or whether people lose unsaved work.
• Size host pools against peak, not average. CPU and memory experiments apply pressure while you watch session responsiveness. A desktop that feels slow generates vague complaints that are hard to diagnose and easy to dismiss.
• Validate your scaling schedule. Most host pools scale on a timetable written during rollout. Testing shows whether that timetable still matches how your organization actually logs in.

For a workload measured in interrupted work rather than error rates, the cost of a bad host lands directly on your colleagues.