Complete Guide to VDI Performance Testing | Citrix, AVD, Omnissa Horizon

Don't guess — measure before you scale.

• Capacity planning — Know how many users your VDI farm can support before you add them.

• Change validation — Prove that OS patches, GPU drivers, or broker updates don't degrade UX.

• Migration confidence — Baseline Citrix, test AVD, compare side-by-side before cutover.

• SLA validation — Document P90/P95 response times and error rates for contracts.

• Read the Citrix → AVD migration checklist

One wizard per platform — not generic templates.

Each wizard guides you through connection, users, agents, display, scenario, and advanced settings. Same phase model (Idle, RampUp, Sustain, RampDown) across all platforms.

Design scenarios that mirror real users.

• Workload selection — Use workloads that match your user personas (Office, browser, CAD, etc.). Configure at /config/workloads.
• Phase timing — RampUp simulates gradual login; Sustain holds load; RampDown logs off. Avoid instant spikes.
• Think time — Add delays between actions. Real users don't click every second.
• Concurrency vs. throughput — More vUsers ≠ better test. Match your target peak concurrency.

Watch tests in real time — live cockpit.

• Active sessions — See running tests at /testing/active
• Per-agent status — Logon, running, error states
• Real-time metrics — Response times, throughput, error counts
• Abort/control — Stop tests early if something goes wrong

The /testing/active cockpit surface. Shown here between runs — the cockpit fills with per-agent status, real-time metrics and abort controls the moment a run starts.

Compare up to 5 tests side-by-side.

• Date range, profile, status filters
• Overlay P90, P95, error rates across runs
• Detail modal: Overview, Moments, Measurements, Errors
• Trend strip for historical analysis

Route: /testing/results

5 mistakes that kill VDI test validity.

1. Wrong agent type — Use VDI agents for session-based load; Full agents for fat-client. Configure at /agents/vdi and /agents/full.
2. Too few agents — One agent per vUser for session-based. Scale agents before vUsers.
3. Ignoring SUT monitoring — Add counter templates and SUT machines. Route: /configuration/systems-under-test.
4. No baseline — Always run a baseline before changes. Compare at /testing/results.
5. Generic profile for Citrix — Use the Citrix wizard. Connection and display settings matter.

Platform-specific wizards, live cockpit, results comparison. No lengthy setup.

• Request a tailored demo
• See Load & Performance Testing
• See LoadGen Agents
• Citrix platform support
• AVD platform support
• Omnissa Horizon platform support
• RDS platform support