Relative Density
What a change costs per host.
Every image update, agent rollout or new application raises the same question: does a host still carry as many sessions as before? Relative Density answers it. Set your own breaking point, for example logon time or a host resource staying over its limit, and LoadGen measures how many sessions ran before that rule fired, per host when it reads your hosts, in this run and in your baseline.
How the measurement works
You set the breaking point
For example logon time, or a host resource staying over its limit. LoadGen measures how many sessions ran before that rule fired.
Per host, where hosts are read
Measured per host when LoadGen reads your hosts, in this run and in the baseline you compare against.
A 90% interval, and a verdict
The result is judged against the density budget you set. A change whose interval straddles your budget reads Inconclusive rather than Over.
Comparability is checked
LoadGen checks that the runs are comparable and says so when they are not.
A ramp preset, and an early stop
A Density ramp preset sets up the ramp for you, and a run can stop early once your breaking point has held.
The verdict
Inconclusive is an answer, not a failure.
The result comes with a 90% interval and a verdict against the density budget you set, so a change whose interval straddles your budget reads Inconclusive instead of Over. LoadGen checks that the runs are comparable and says so when they are not.
A Density ramp preset sets up the ramp for you, and a run can stop early once your breaking point has held.
Capacity planning with LoadGenBefore an image update
Measure what the new gold image costs per host before it reaches a single user.
Before an agent rollout
A security or management agent added to every desktop has a price in sessions. This is how you read it.
Before a new application
Adding an application to the image changes what a host carries. Measure it against the baseline you trust.
What density does not tell you.
- Density is measured on the hosts and image you tested. It is evidence for a go or no go on that change, not a forecast for other hardware or for production.
- It is not a measure of the LoadGen Score. The score answers how the run went; density answers what a change costs per host.
- Per-host density needs LoadGen to read your hosts. Azure Virtual Desktop session hosts are read through an Azure Monitor connection.
Measure the change before it reaches your users.
Bring the image you are about to ship and the baseline you trust. We run both and show you what the change costs per host.
