Azure Virtual Desktop | Automated scaling, imaging & monitoring

QUICK LINKS:

00:00 — Introduction

00:23 — Scaling plans

03:29 — Personal scaling plans

04:30 — Up-to-date VM images

07:13 — Azure Virtual Desktop Insights at Scale

09:02 — Session History chart

10:06 — Wrap up

Link References:

Check out our complete playlist at https://aka.ms/AVDMechanicsSeries

Unfamiliar with Microsoft Mechanics?

As Microsoft’s official video series for IT, you can watch and share valuable content and demos of current and upcoming tech from the people who build it at Microsoft.

• Subscribe to our YouTube: https://www.youtube.com/c/MicrosoftMechanicsSeries
• Talk with other IT Pros, join us on the Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog
• Watch or listen from anywhere, subscribe to our podcast: https://microsoftmechanics.libsyn.com/podcast

Keep getting this insider knowledge, join us on social:

• Follow us on Twitter: https://twitter.com/MSFTMechanics
• Share knowledge on LinkedIn: https://www.linkedin.com/company/microsoft-mechanics/
• Enjoy us on Instagram: https://www.instagram.com/msftmechanics/
• Loosen up with us on TikTok: https://www.tiktok.com/@msftmechanics

Video Transcript:

-Today, I’ll walk you through what you need to know for running your Azure Virtual Desktop infrastructure efficiently, including scaling plans to drive up utilization and save costs, how to create and manage images, as well as your options for monitoring and gaining insights on the service to take action. This is part four in our series on the service, which you can find at aka.ms/AVDMechanicsSeries.

-Let’s start with making the service as efficient as possible using auto scaling. Scaling plans are one of the primary ways to configure Azure Virtual Desktop for cost savings. This capability is most effective when using multi-session versions of Windows client and Windows Server, with several users simultaneously signed into the same host. Now using scaling plans, you can scale your session host VMs in a host pool up or down automatically, so that you aren’t paying for idle resources. To create and configure scaling plans, you’ll start in the Azure Virtual Desktop Scaling plans page and hit create scaling plan. In basics, you’ll define the normal Azure parameters with subscription and resource group, a name, location, and importantly because these are triggered on schedules, you’ll confirm your time zone for this scaling plan.

-Next, in schedules you’ll define the timing for scaling plan actions. Here you can define multiple schedules. For example, if you want different scaling rules in place for weekdays versus weekends. So I’ll add a schedule for weekdays and as I expand the repeat on control, you’ll see I can select which days of the week. I’ll stick with the default Monday to Friday range. Then in ramp-up this is where you define when hosts start up. For example, if people start to sign in at 9:00 a.m. local time, I can keep this value of 8:00 a.m. to ensure hosts are started and warm just prior to them signing in. In the load balancing algorithm, I can choose from two options. If I select breadth, it will evenly distribute users across available session host VMs, or conversely depth first will assign users to the host with the highest number of active user sessions until it reaches the max session limit.

-Next, you’ll define the minimum percentage of hosts. So this represents how many at a minimum will be available or warm at any time. Then the capacity threshold which determines when additional hosts will be allocated once the threshold is reached. For example, if a host pool supports 100 active user sessions, and you’ve set your threshold to 60%, once the 61st user signs in another session host will be turned on, so it can be ready for additional users to be allocated to it. In peak hours, you’ll define your peak start time, so I’ll keep 9:00 a.m. And I have the flexibility to change the load balancing algorithm here as well during operational run state, and I’ll share a tip on that in the next tab. Then in ramp-down, you’ll reverse your steps from before, except here, you can also force logoff to drain your host VMs and shut them down to save costs as the schedule comes to a close.

-And a tip here is to change the load balancing algorithm from breadth first to depth first. You may begin the day using breadth first to make sure hosts are ready to be signed into in the morning and keep that option during peak usage. Then move to depth first for ramp-down and off-peak hours to reduce costs in low utilization periods. Here you can also define the delay time before logging out users, and the notification message that’s displayed to your users prior to logging them out. And finally, in Off-peak hours, you’ll specify the timing and load balancing algorithm there as well.

-Now, we’re also rolling out a new option that allows personal dedicated hosts to use scaling plans. These are also configured as part of scaling plan creation and you’ll add schedules in the same way too. Personal scaling plans allow VMs to start up automatically on a schedule so the user can sign into a warm and running host. Optionally, if a user tries to access a personal host that is not already up and running, if you’ve enabled Start VM on Connect, the user can still sign in, but it’s going to take a little bit more time to get to a session, since it will need to start the VM first. And importantly, you can automatically shut down and deallocate VMs on a schedule when not in use, or based on the parameters you set. And beyond scaling, you can also take advantage of the Azure Savings Plan, as well as reserved VM instances where committing to long-term use can bring you further cost reductions. Now, one more related pro tip here is to apply tagging as you create resources and even apply them against running resources using policy, in order to easily track your related spend in Microsoft Cost Management.

-Now, another primary consideration when operating your Azure Virtual Desktop infrastructure is how you keep your VM images current with required software and configurations, in addition to Windows updates. Here the best practice is to have a monthly process to layer on your customizations in a repeatable and templated way using a new fully up-to-date Windows VM image. In Azure Virtual Desktop, you’ll start by using a custom image template that defines your image. In basics, you’ll add the usual parameters and your managed identity. In source image, you can choose a platform image from the marketplace, a managed image, or one from the Azure Compute Gallery. So I’ll choose a platform image in my case, supporting multi-session and including Microsoft 365 apps.

-In distribution targets you can manage where the resulting image will be saved to, either as a managed image, or you have the option to specify Azure Compute Gallery, where you can also determine the location where images are replicated globally. So here you’ll also define image and output names, which will help you later as you search for images. In build properties, you can define the timeout, VM size, disk, group, and network configs. Then in customizations, there are a number of built-in options you can leverage, including installing language packs, where there are a dozens to choose from. And here I’ll add Dutch and there we go.

-Next, I’ll define my FSLogix installation and configuration for user profile management. So I’ll choose my version. I’ll choose a few more scripts here, one for RDP Shortpath and another for Screen Capture Protection. Then you’ll see Microsoft Teams optimizations with client, as well as WebRTC, and Visual C++ Redistributable requirements to improve voice and video in remote sessions. There are also options for session timeouts, or configure time limits for disconnected sessions, idle sessions, active sessions, and time to log off. And I’ll add multimedia redirect configurations with all of their prerequisites in these dropdowns. Also, if you want to, you can define application scripts for Appx packages and Office. In addition to options for timezone redirection and installation of any out of band windows updates not already in the monthly image.

-Then beyond what’s built-in, you can optionally include your own custom PowerShell scripts that can run on top of the base image with your app installations and other configurations, like the one I’m choosing here. It just needs a reachable URI path for your script. And as usual, you can add tags. And then you can make sure everything looks good and create your template. Once configured, you’ll see your template here. And by selecting the template, you’ll be able to start the image build, which sends it to the Azure Image Builder for completion. The Azure Image Builder service then distributes the resulting image with everything specified in your build to the locations defined in the template. So now with best practices in place for scaling and image management, let’s take a look at how you can get an end-to-end view of how your services are running using Azure Virtual Desktop Insights at scale.

-We’ve recently added the capability to see all of your host pools and subscriptions in one view. You can find an entire dashboard of built-in Azure Monitor Workbooks, which starts with an overview of your environment with its total capacity to view all services deployed. Connection diagnostics to see whether users can connect easily to resources. Connection performance, to monitor time to connect in near real-time. And utilization, which is key to keeping control over costs and helping avoid idle resources. And you can drill into these areas for even more fidelity. For example, in Connection Diagnostics, you can quickly see the volume of connections to your hosts. And by toggling focus on failures, you get a nice view to help troubleshoot by session hosts, corresponding users, resources, or apps, and even RDP client types to help pinpoint issues. Then in Connection Performance, you’ll see one of the most important proof points for quality of service, the round trip times, which are a great measure of network latency for users as they interact with hosts. Typically, a good round trip time is under about 140 milliseconds. So identifying higher round trip times and fixing them is key to better user satisfaction and making remote sessions feel local.

-Now, another key metric to focus on and a big part of keeping costs under control is utilization. Here you can see a summary of sessions to host pools to see how they’re being used, time of day usage patterns, the max number of users per core, available sessions over time, your active session host count, along with idle hosts, and finally, monthly active user information.

-Now, if we bring this back to scaling plans, there are a few specific things you can do with utilization insights to inform them. So here you can see the session history chart. You’ll use this view, for example, to identify peaks in session count and also the times these peaks occur. And if you spot repeated patterns, you can use these values as you create minimum and maximum thresholds in your scaling plans, and determine when to turn off idle resources. The blue line represents the total number of active sessions, and the red line is a subset of that with idle sessions. So here you can see that idle sessions are proportionally high, so you might want to enforce log offs using policy to reduce this proportion of idle sessions.

-Something else to adjust for if you see it are sustained peaks in idle session hosts in the session host count charts. This means that they’re not being utilized and some can be turned off during these times. So here you’re seeing one on a weekend when we can ramp-down. Taking a data-driven approach to scaling will ultimately optimize your costs and help achieve the most efficient resource utilization with Azure Virtual Desktop.

-So these are just a few best practice recommendations for operating and keeping your Azure Virtual Desktop optimized, up-to-date, and well utilized. And again, this is the fourth in our series on mechanics for Azure Virtual Desktop. So check out the complete playlist where we cover setup and your configuration options at aka.ms/AVDMechanicsSeries. Be sure to subscribe to our channel for future updates. And thanks for watching.