Welcome to this module on implementing autoscaling in host pools. In this module we will cover the following topics:

• Azure Virtual Desktop Scaling plans
• Requirements for auto-scaling in AVD
• A demo where I will configure an auto-scaling plan for Azure Virtual Desktop

Let’s start off by discussing scaling plans.  Traditionally, scaling was implemented within Azure using either Powershell scripts or Azure Automation, however, you can utilize a feature called scaling plans.

Scaling plans allows you to define ramp-up hours which is the time you want to increase resources when the usage increases.  

Peak hours which is the time of day when usage is at its highest.

Ramp-down hours when usage starts to drop off.

You can then configure associated triggers which will kick off your scaling plan if those triggers are met, for example, a certain percentage of CPU usage.  Finally, your scaling plan must include an associated schedule for at least one day of the week.

Let’s now move on to discussing requirements for autoscaling in Azure Virtual Desktop.  There are two main items that need to be in place before you can start implementing autoscaling.  The first is that you need to create a custom role-based access control, better known as RBAC, role.  Once this role has been created, the second item that needs to be in place is that you need to assign this role to the Azure Virtual Desktop application.  

Once these two requirements are fulfilled, you can start configuring autoscaling in Azure Virtual Desktop

To finish off this module we are going to complete a demo of implementing a scaling plan in Azure that you can utilize with your AVD environment

Here we are in the Azure Portal.  We need to navigate to the scaling plans section of the portal. We do not currently have any existing scaling plans, so let’s select ‘Create’.  We now need to ensure we select the relevant subscription, resource group and give the scaling plan a name. You will need to be careful when selecting the location as not all locations support scaling plans.  In our example, I am selecting Canada East as I know this region is supported.  Finally, I am giving it a friendly name and moving on to the schedule plan.

We need to select ‘Add schedule’.  We now have a number of different panes we need to complete.  First, let’s enter general information like the schedule name.  We then need to select which days we want the schedule to repeat on, in our case, we select each weekday.

For ‘ramp-up’ hours, the repeat on days and time zone are already entered by default, so we need to enter the start time.  Please note this utilizes the 24-hour clock system.  We then need to specify the load balancing algorithm.  I am going to leave this on bread-first.  We then need to specify the minimum percentage of session host virtual machines to start for ramp-up and peak hours.  I will leave this on the default of 20%.  Finally, on this pane, we need to enter the percentage of used host pool capacity that will be considered to evaluate whether to turn on or off virtual machines during ramp-up and peak hours.

If we now move onto the ‘Peak Hours’ pane, again the days and time zone are already selected and we need to specify the start time and load balancing algorithm which I am going to change to bread-first.  The threshold capacity is grayed out and set to 60%, therefore we cannot edit this.

Now we move on to the ‘ramp-down’ pane.  The first value we need to set is the start time, then the load balancing algorithm, which I am again going to set as breadth-first.  We now need to set the minimum percentage of session host virtual machines that you would like to get to for ramp-down and off-peak hours.

Next, we need to select the percentage of used host pool capacity that will be considered to evaluate whether to turn on or off virtual machines during ramp-down and off-peak hours.  We can then specify if we want to force users to log off or not, set a delay time before forcing the users to log off and customize a notification message.

The final pane to complete the schedule configuration is the ‘off-peak hours’ pane.  Here we need to specify the start time, and load balancing algorithm.  Again I am changing this to bread-first.  Please note we cannot edit the capacity threshold again.

Now the schedule is configured we need to select the host pool we want this to apply to.  In our case, we do not have an active host pool in this tenant. Finally, we can enter any tags we want to set, before reviewing and creating our scaling plan.