AZ-140 8 - Final Thoughts on Azure Virtual Desktops

This information isn't meant to be a complete resource for AZ-140 certification.  Just simply a quick primer on setting up virtual workstations.  Before deploying Azure Virtual Desktops, there are several other resources that you would need to know about and likely another 8-16 hours of training and labs.  Many people out there have done a great job with paid training for these things, including Thomas Mitchel, who has the best training out there in terms of price vs. information.  His work may be found here:

You'll want to learn about things like FSLogix for roaming user data between computing sessions, reducing sign-in times, optimizing IO, providing local profile experiences, simplifying management of images, and providing Java for applications (hopefully, this need eventually disappears!).  More info on FSLogix can be found here:

My original goal here was to determine if remote desktops in Azure were hype in these times of remote or hybrid work or not.  I'm not sure that I've answered that question for myself at this point.  I think there are many positives and several negatives (I know Dr. Witz, be specific but, I can’t because I don’t feel like I know them all!).  From a positives perspective, it's nice that there is no need for individual software deployments with manual installs or deployments using SCCM/Matrix42/LanDesk or other software asset management tools (SAM).  Beyond that, there's no need to image systems, deploy hardware, perform domain joins on systems, worry about remote GPO updates, very quickly scale up new sessions, backup profiles, etc. more pronounced and often touted advantages.  I see the containerization of the system as a positive for security; you can maintain clean gold images without worrying as much about what a user is up to or what their home environment looks like.

As a con, it's damn expensive.  Excuse the expletive.  I had a $200 credit, and I was through that credit with $19 left at the end of this in two weeks, and I kept resources shut down while I wasn't working in the environment.  This could shoot up wildly with a full-blown enterprise deployment with thousands or tens of thousands of users.  You would need to map out the cost savings of not providing hardware, auto-provisioning applications, and all the costs of supporting thousands of personal systems.  I chose what I felt were the lowest viable hardware specs also for this setup.  Looking back through screenshots I’ve created for this write-up, a typical workstation would probably be upwards of $189 a month, just for the workstation component.  This doesn't include the compute on the server infrastructure, storage, Azure, etc.  That's $2268+ a year just for the virtual desktop virtualized hardware. 

I don't know about your infrastructure but, we try to stretch laptop/desktop hardware 3-5 years on average.  NOW, I realize that multiple users can leverage a single virtual desktop.  There is some balancing act in determining how many folks you can provision to a workstation before you run out of resources and impact user performance.  I don't know where that number/limit is, and it likely depends on your users' compute load and your acquisition of RDS CALs for Windows 10.  Sharma (2020) discusses the concept of auto-scaling Azure Virtual Desktops to save on costs at the following link:  While using Euro's, Lucas Clara (2020) provides a great summary of how many users you could theoretically host per VM here: 

As another con, I'm not a massive fan of supporting personal equipment.  There are too many exponential combinations of what you could run into from software installed on individual systems (think of the number of AV/Firewall/IPS/EDR vendors that you could run into alone).  Of course, given the resources, personal support would be fine but, resources are often tight for support.  You could make a case for deploying or reusing internal hardware to expand the lifetime and simply leveraging virtual desktops, or booting a free Linux image to load the virtual desktop environment in a browser, etc.  However, let's take those cases out and talk about supporting BYOD in a fully remote/hybrid work environment. We are back to our service desk(s) supporting home internet connectivity, client issues, browser issues, clipboard/printer mapping, etc.  We run into user complaints if we break personal hardware, etc.  You could publish guidance on minimum requirements and NOT support personal hardware, but this feels a bit unreasonable in a remote/hybrid environment.  You would likely need to work with your legal department on drafting language for liability on damaging personal hardware in the home office space, set limits on what your team is willing to do on a personal system, or simply accept the risk of breaking personal hardware/software. 

Another consideration, although not a con, another consideration is that this isn't for you if you're not willing to plan.  Naming conventions are key; careful planning is key.  I would hate to see an environment where you slap random names on things; it could get real confusing, real quick.  Come in with a plan, maybe engage with an outside vendor. 

Long story short, this would probably be a good case study for a DBA student out there on the viability of virtual desktop infrastructure for small/medium/large businesses.  I don't have time to look at a ton of variables in two weeks using only personal time, and most of what I've found is slanted from a vendor perspective.  I'm sure the application and desktop virtualization organizations out there would tell you that there are substantial cost savings and I've certainly used the technology at various organizations.  I'm just not sure the argument is there to switch all of your remote and hybrid workers to it.  I think it would make for a good paper. 


Sharma, N. (2020). Azure Windows Virtual Desktop: Cut infrastructure costs with auto-scaling.

Related Articles

Free Joomla templates by Ltheme