Peak Load · /// Spike Validation
Model the peak before the calendar lands on it.
Finance-close, year-end, open-enrolment, e-commerce peak. Predictable calendar events that production discovers too late. LoadGen spike simulation models the peak with measured concurrency across Citrix HDX, AVD ARM, Horizon, RDS, Web. Measured peak vs modelled peak. Close the gap before users do.
Spike + soak + warm-up + cool-downCross-vendor (Citrix · AVD · Horizon · RDS · Web)Measured peak, not estimated
Spike simulation. Measured peak vs modelled peak.
The Problem
Calendar peaks beat steady-state assumptions.
Most VDI estates are sized for steady-state load. The peak event. Finance-close Monday morning, year-end open-enrolment, Black-Friday e-commerce surge. Exposes the gap. Production discovers it; spreadsheet sizing didn’t.
Steady-state hides the peak.
Capacity dashboards report steady-state averages. Peaks live in the percentiles. p95 and p99 latency. That only appear under deliberate spike load.
Spreadsheet models miss the cliff.
Linear capacity models assume gradual scaling. Real peaks are non-linear. Host pools, brokers, and shared storage break at threshold counts not visible in linear extrapolation.
Multi-stack peaks need cross-vendor framing.
A finance-close peak crosses Citrix (legacy), AVD (newer apps), and Web (portal). One peak event, three test plans. Unless the scenario engine spans all three.
Why LoadGen for peak validation
One scenario shape. Spike phases. Cross-vendor.
The same scenario engine that runs steady-state load tests adds deliberate spike phases. Warm-up, steady, spike, cool-down. Cross-vendor: Citrix HDX, AVD ARM, Horizon, RDS, Web. Measured peak surfaces before the calendar event.
Spike phase profiles
Configurable warm-up / steady / spike / cool-down phases. Spike-shape profiles model the actual peak event. Finance-close burst, open-enrolment ramp, Black-Friday surge.
Per-step latency at peak
Per-step latency, p95 / p99, host-pool density. Visible as the spike progresses. The component that breaks first surfaces before production does.
Cross-vendor on one timeline
Citrix HDX peak, AVD ARM peak, Web portal peak. Captured on the same scenario engine, overlaid on one chart for direct comparison.
Spike orchestration
Warm-up → steady → spike → cool-down, live.
The cockpit during a spike run. vUsers ramping into the peak, per-step latency growing under the spike. The gap between modelled peak and measured peak closes in front of you.
- vUsers ramp across warm-up / steady / spike / cool-down. Visible as the run progresses.
- Per-step latency + p95 / p99 at peak.
- Modelled peak vs measured peak gap closes in front of you.
- Spike data feeds straight into capacity sign-off for the calendar event.
Live orchestration cockpit during a spike run.
Capability
What ships with peak-load validation.
Configurable spike profiles
Warm-up / steady / spike / cool-down phases authored in the load-profile wizard. Spike shape, duration, and concurrency tuned per scenario.
Learn moreLive cockpit at peak
Per-step latency, p95 and p99 visible as the spike progresses.
Learn moreSUT Monitoring during spikes
Citrix Broker, AVD connection-broker, Horizon Connection Server, RDS Connection Broker counters bound to every spike test. Infra correlation at peak.
Learn moreMulti-test comparison
Up to 5 spike runs overlaid. Week-over-week regression, modelled vs measured, pilot vs production scale.
Learn moreCross-vendor scenario reuse
The same .lgs scenario authored once spikes across Citrix, AVD, Horizon, RDS, Web. One peak event, one scenario.
Learn moreEvidence for capacity sign-off
Every spike run captured in queryable history. Exportable to PDF / JSON for the peak-event readiness review.
Learn moreCross-vendor overlay
Citrix peak vs AVD peak vs Web peak. On one chart.
Multi-test overlay compares spike runs across stacks. Same scenario shape, three stacks, one chart. The cross-vendor peak picture in one view.
- Same .lgs scenario spikes across Citrix HDX, AVD ARM, Web. Up to 5 runs overlaid.
- Drill into Moments, Errors, per-step deltas at peak.
- Identify which stack breaks first under the modelled peak.
- Cross-vendor peak readiness in one chart, not three spreadsheets.
Multi-test overlay. Cross-vendor spike comparison.
Outcomes
What peak validation looks like in numbers.
Before
Spreadsheet
After
Measured
Before
Multi-tool
After
One chart
Before
8
After
1
Before
Weeks
After
Days
Where peak validation lives
Three calendar events worth measuring.
Finance-close Monday morning
Monday-morning logon storm + finance-app load. Spike profile models the simultaneous arrival of every controller / accountant / ops user. Capacity sized for the burst, not the average.
See use caseYear-end / open-enrolment
Predictable calendar events with disproportionate load. Multi-week spike validation models the ramp-up across the event window. Measured before users hit the calendar.
See use caseE-commerce / Black-Friday
Web portal spike testing alongside Citrix back-office and AVD operations staff. Cross-vendor peak readiness on one scenario engine.
See use caseValidate your peak before it happens.
We’ll author a spike scenario in the wizard on a call, fire it from Full or VDI agents, and show you measured peak across Citrix / AVD / Horizon / RDS / Web. Live, before the calendar event.
Questions
Frequently asked.
How does peak validation differ from steady-state load testing?
Steady-state load tests model a constant level of concurrent users. Peak validation deliberately spikes the load. Warm-up / steady / spike / cool-down. To surface non-linear breakpoints (broker queue saturation, shared-storage contention) that steady-state misses.
Can the spike profile model real calendar events?
Yes. The load-profile wizard configures phase shape, duration, and concurrency. Customer scenarios commonly include Monday-morning finance-close (sharp 30-minute spike), year-end (multi-day ramp), and e-commerce flash sales (5-minute peak).
Does this work for multi-stack peaks (Citrix + AVD + Web)?
Yes. The same .lgs scenario engine drives Citrix HDX, AVD ARM, Horizon, RDS, and Web. One peak event, one scenario authored in the wizard, replayed across all stacks. The multi-test overlay shows the cross-vendor peak picture on one chart.
How accurate is the modelled peak vs the real peak?
Published target for host-pool right-sizing is ±5% accuracy on AVD, compared against ±40% typical of spreadsheet sizing models (source: /platforms/avd OUTCOMES). Numbers come from per-host measured concurrency, not modelled assumptions.
What does peak validation cost?
Load Testing module is €1,099 per week at the 50-vUser tier, scaling to 25,000 vUsers. Multi-week terms run from 1 week to 5 years. Peak-event engagements often combine Load Testing with End-to-End Monitoring (€899/Agent/mo) for SLA tracking through the event window.
Can LoadGen Services manage the peak validation for us?
Yes. LoadGen Services (consulting + managed runs) is available where in-house bandwidth is tight. Particularly common for high-stakes calendar events (year-end close, regulatory reporting peak).
