Why Microsoft hates VDI

For years (even decades), Microsoft hasn't really had to market the Windows platform. It was a total monopoly in the business world and just about every new business app was written for it.

For years (even decades), Microsoft hasn't really had to market the Windows platform. It was a total monopoly in the business world and just about every new business app was written for it. One of the classic moves we see from monopolies (across all industries) is that when new ideas come along that disrupt their dominance, the monopoly will ignore, deny, lash out, and eventually try to copy the new way of doing things. This is exactly what we're seeing with regards to Microsoft and VDI.

For the longest time, Microsoft was in complete denial that VDI existed. Finally they wised up and started to talk about VDI, although typically comparing it to traditional desktops or Remote Desktop Sessions / Terminal Server. (Here's their latest paper showing the VDI is more expensive than traditional PCs.) Then when they finally did decide to legally allow customers to use Windows for VDI, they did so with a completely ludicrous and arcane licensing model that few customers could effectively use. And their own VDI offering was created as an afterthought add-on that recycled all the existing remote desktop components and recombined them into a VDI "Suite" aimed for "low to medium complexity environments."

The problem came from the fact that Microsoft's sworn enemy VMware was leading the charge towards VDI. VDI shares many architectural similarities to Microsoft Remote Desktop Session Host, so Microsoft couldn't pooh-pooh the concept entirely. But there was never any danger that Remote Desktop Session Host would negatively affect the Windows client business, as no one truly believed that any significant percentage of Windows dekstop users would fully migrate to remote sessions.

So VMware pushed VDI. Microsoft pushed back.

"But that's all changed now!" some people will try to convince you. "Now Microsoft loves VDI!" People who believe this cite the fact that Microsoft has dropped the additional cost of the hated VECD license and that their new RemoteFX protocol promises to deliver a great user experience via VDI.

To both of these I respond, "Balderdash!"

If Microsoft wanted to make VDI easy to use, they would have allowed any version of Windows 7 to be accessed via VDI instead of today's rules that state a client device must be enrolled in Software Assurance (SA) to reap the benefits of VDI. (And the additional tools like App-V would be included in the OS instead of an SA-only add-in option.) Thin clients that don't qualify to be covered by SA are now required to have their own special license called "VDA" in order to use VDI, and this VDA license costs a whopping $100 per device, per year! So while SA only costs $100 the first year and then drops to about $40/year for renewal, the VDA license for a thin client costs $100 year after year after year. Does a company that charges a $60 per year thin client device tax sound like a company that loves VDI?

And then there's RemoteFX. Was this protocol created because Microsoft loves VDI? Hardly. RemoteFX exists for two reasons: First, it's a ploy to push Hyper-V (since RemoteFX requires Windows 7 VMs to run on Hyper-V. It doesn't work with VDI environments based on VMware vSphere or Citrix XenServer). Second, Microsoft's push for RemoteFX decoders in things like LG televisions shows Microsoft's intentions with RemoteFX go beyond creating it as a nice gift for the VDI market. RemoteFX is part of something much bigger. (Xbox in the cloud, anyone?)

[UPDATE] Commenter Mads Sorenson also points out that there is no Service Provider License Agreement (SPLA) for the Windows client. So if you want to provide Windows desktops as a service to customers, the customer has to own the license (which means the customer needs SA or VDA). This makes all of that more complex. If Microsoft really wanted to encourage Windows desktops as a service, they'd offer a SPLA for desktops like they do for Remote Desktop Session Host, Office, etc.

So you can see it's easy to find lots of anecdotal stories that suggest that Microsoft doesn't like VDI. But why not?

There are a few reasons for this. The first thing to remember is that the Windows desktop monopoly is under attack on multiple fronts. Of course we all know about Apple in the consumer space. And we know about iPad and Android tablets. And of course new app developers have the choice to write their apps for Windows or the web or Java or as rich internet apps... So the problem is that a lot of new apps today aren't written specifically to be Windows apps. (Or if they are, there are also native Mac, iOS, and Android apps written at the same time.)

Over time we'll see fewer and fewer traditional Windows apps. Then at some point, people will realize they don't need Microsoft Windows on their client device just to run a few Windows apps. And they certainly don't need Windows (or any "full" OS.. Mac, Linux, etc.) to run web apps. So at this point it will be easy to migrate the few remaining Windows apps from the client into the cloud. Then they can be delivered back to the client as seamless RemoteApps (maybe even via RemoteFX).

At this point you might be thinking, "Ok... so what? Why does Microsoft care? If people are still using Windows, who cares if it's running in a remote datacenter versus running locally?"

The problem with the only Windows apps being remote VDI apps is it sets a bad precedent. Microsoft has spent a lot of R&D creating the "richness" of the Windows user experience. But if we strip that off the client and just deliver a seamless application via some remoting protocol, the nuance and art of the richness is lost and Windows is emasculated. Why do I need Windows 7 just to deliver my stupid old Windows app? (Look at all the new features in Windows 7. Which of these are applicable if you're just using accessing a single seamless RemoteApp?) And if that's the case, why do I need Windows 8 for VDI? Why do I need SA? I can just use the same copy of Windows that I bought ten years ago since it's just a single old remote app -- it's not the client and it's not affecting the overall client user experience). This is what I meant when I wrote that Windows itself is in danger of becoming the "XP Mode" of the cloud.

Microsoft has worked hard over the past twenty-five years to build the Windows franchise and create demand for it. It's not going anywhere anytime soon, but it's clear that Microsoft benefits from Windows being installed locally on client devices. VDI challenges that.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Also, don't forget that no Service Provider can legally provide Windows 7 in the cloud, because this is not allowed under the SPLA program, see the bottom of page 3: https://bit.ly/ii5Gug

I am a hoster who wants to provide Windows-based desktops as a hosted service. Do my customers need to pay for Windows VDA? OR

Is there a Service Provider Licensing Agreement (SPLA) for Windows VDA so that hosters can provide Windows-based desktops as a service to third parties?

Currently, there is no SPLA model for Windows VDA. Hence, customers who subscribe to desktops from a third-party hoster will need to pay Microsoft for a Windows VDA license for each device accessing Windows client virtual machines in the datacenter. Additionally, hosters need to ensure that they isolate the hardware and other resources for each company (i.e. no two customers can share the same set of resources, such as hardware, storage, etc).


Brian can you post stats on what browser is used to access your site the most? On Crunch Chrome leads the pack with that cow IE in distant 4th. I know it's a little bit off the topic but I am just wondering. Thx



You are actually describing why Microsoft would hate Application Virtualization because it allows Windows apps to be run on non-windows client devices... Since VDI uses Windows in the datacenter, as long as Windows is used I doubt MS cares.

IMO, MS doesn't hate/love VDI, they are just trying to find a way to get the most profit from it without making it impossible. The current SA or VDA licensing scheme is much like their RDS CAL, albiet much more expensive but I think it is to limit adoption and wait for a broader hyper-v acceptance.

MS shouldn't even care about investing in their VDI Suite, after all they have Citrix for that extension.

MS should just concentrate on further enhancing the Windows User Experience and Hyper-V/RemoteFX. After all, the only reason Hyper-V exists is because Citrix XenServer couldn't beat VMware on it's own.

Now we are seeing VMware folding it's cards and offering vSphere support for Hyper-V, which shows acceptance.


I wonder if Microsoft would legally be allowed to say do something like, "If you use Hyper-V, you can run as many Windows VMs for free as you want." Or if they'd be allowed to have SPLA for Windows 7 but only if it ran on Hyper-V?

I'm guessing probably not? (Even if it was legal, everything would think a move like that was not nice.)


@system.fracture - I'm not sure of the numbers exactly since I looked at them a week or two ago, but it's still overwhelmingly IE. FF is a distant second, with Chrome right on its heals.


@Icelus if vSphere is now managing Hyper-v I would not be surprised to see Systems Center manage more platforms in the future. The world will certainly have more variety but legacy Windows is going nowhere fast. I'll predict right now that in 2020 the majority of the world 80% will still use the legacy desktop in the enterprise on Windows 7. Sure more people will use cloud something but change will take a long time to go mainstream. MS understands that and will FUD the market to slow the change down while they try to figure out what to do next. The greatest danger they face is themselves and a Windows only world. I still don't get why they don't just buy Citrix and use them as a division to have a broad play to embrace devices etc. I guess the bet MS is making is that they can convince developers to move to Azure. They want a single OS on a single dev platform. That's where I think the battle will be won or lost. Developers, developers developers......as the famous Steve Ballmer video on youtube illustrates.


Good insights. The biggest risk that Microsoft runs is already a well-advanced trend: corporations will stop building OS-dependent apps at all because the desktop is a bear to manage (of course, MS has t$$ls for that), OS API changes (XP to Vista for example) can render costly customizations useless, and as you document here, MS is making centralized desktops through VDI so costly. With HTML5 and other technologies, the browser becomes the desktop. At that point, the least expensive device wins the user hardware race. It doesn't need many features--the features live in the cloud. Users get centralize app management, instant app updates, and the OS doesn't matter.



I think you hit the nail on the head, perhaps not explicitly. VDI (and related technologies) is an enabling "bridge" technology to let organizations take advantage of trends like consumer IT and the vast array of new client access devices. The bridge leads to a world where apps no longer rely on a Windows desktop to run. Its truly why I think, for the first time in the 20+ years I've been working with Windows client, it is as close to becoming irrelevant as I've ever seen.


Microsoft makes money and shareholders are happy. F* the customer after a certain point.

What’s really silly is that Apple doesn’t allow this either. Imagine accessing a fully functioning Mac desktop in the cloud from your iDevice. (I understand the “app in the cloud” argument).

What it comes down to is that companies either follow the www.icloud.com route or have to host their own private clouds for HDV/VDI.

The moment this revenue model gets threatened (which IS happening with Google Chrome), Microsoft will begin to offer versions of their desktop OS specifically designed for HDV/VDI.


Gabe - thanks for responding! are you kidding?

What is everyone doing using IE to browse the web?? :-p


Brian, in reference to your comment/question about M$ licensing VM's for free once you have HyperV, they're already doing that with server OS, so why wouldnt they with VM's?

Here's an excerpt on W2K8 R2 DC version.

Windows Server 2008 R2 Datacenter enables you to run any number of instances of the software on a server.. Each server must be licensed according to how many physical processors are in the server. (Virtual processors are not used for the licensing of Windows Server 2008 R2 Datacenter) After you have determined the number of licenses required, you must assign those licenses to the server, at which point you may run the following at any one time:

• One instance of the server software in the physical OSE, and

• Any number of instances of the server software in virtual OSEs


Its wel known that Microsoft hate VDI, same as few years ago webapps was delema for Microsoft but as GoogleApps is growing, Microsoft came with BPOS and now office365. Same thing is goingon with VDI. Why Microsoft introduce remoteFX? MS know that in the future VDI will be spreding as fire and they do not have VDI product! MS preparing to land in VDI but slowly. Microsoft has basic of VDI, They have hypervisor Hyper-V with dynamic memory, They have remote protocole remote FX and they have application virtualisation App-v. As more and more vendor comming with their VDI Orcale, Linux etc...Microsoft may be also comming with their VDI solution.  


There is one real obstacle to VDI: WINDOWS. The OS is already legacy and behind times. It is unwieldy and really a relic from its height 10 years ago. "VDI" will NOT take off until there is a better solution. Windows does not and will not cut it. There I said it. flame if you want but you know I am right.


With many discussions like this, we have a tendency to view the future as an "all or nothing" proposition. The reality is that a mix of server- and client-side deployment models is here to stay. That said, it would be a fool's errand for Microsoft to try to cling to a pure client-side model, and I think they realize that.

A better approach is to protect the core business to the extent possible (hopefully by continuing to invest in client-optimized features in addition to pulling licensing levers) while investing in new areas like cloud-apps (Office 365), management in the cloud (InTune), delivery methods for traditional apps (App-V, RemoteFX).

Sure, their relative piece of the endpoint pie can still get smaller as new types of devices emerge, but it is also quite possible that they can make it up as value shifts from the endpoint to the cloud. Regardless of the platform, the most interesting and use full apps for me are ones that leveage both local and cloud-based resources.

While as a user I may not like the fact that MSFT is making licensing decisions that get in the way of what I may be trying to accomplish, I am not sure you can fault them for that. It's also really not any different than what Apple and Google are doing from their own positions of strength, case-in-point being the recent Apple subscriptions bruhaha.

Doug Lane



Doug Lane... AppSense?


Yes, settling into some new surroundings in New York this week. Don't worry, still $69.99999999 million left after hiring me. ;)

I'll still defend the honor of client hypervisors as required, though.


@DougLane congrats on a smart move. Client HV honor yes, ability for small startup to really build the core technology and management very hard.....


@DougLane -- congrats as well - you have a great perspective on many aspects of virtualization, including the management piece and personalisation necessities. Whomever replaces you at VC has VERY large shoes to fills. If they haven't hired anyone, ping me, as i know some qualified candidates who are like-minded to Dan M.



If you recall, it took Microsoft a while to sort out the whole Terminal Services license model as well.  When NT4TSE was first released, there was no TS CAL - they tried to force you to buy an NT Workstation license for every client device.  Users pushed back so hard that they had to come up with a different license model, and eventually arrived at one that may not be perfect, but is at least manageable.  My hope is that the same will happen with VDI.

The most annoying aspect, in my opinion, is that the VDA license is only available as a subscription license that, as you point out, costs you $100/year forever.  Simply allowing customers to purchase a perpetual license would be a huge step in the right direction.


This single issue is a very big deal for the industry and I'm grateful to you for sounding the alarm on it.  Here's a related thread on technet:


Also, for those wondering, let it be clear and simple: there is no licensing loophole on this.  If you run a Hypervisor, and provide access to virtual machines running Windows 7 to EXTERNAL CLIENTS using any mechanism, you're in violation of SPLA.  

See scenario 5: download.microsoft.com/.../Windows_Hyper_V_Licensing_Whitepaper_v2_0.docx

Forget about trying a licensing model to comply with, it's expressly forbidden. If you just go ahead and do it, you're reducing the need for resolution by working around or ignoring the problem.

For us all to move forward with this business model, Microsoft must change the policy.  I want to believe it's just a matter of letting the public demand pile up to critical mass until Microsoft releases an updated .doc file allowing this. Maybe I'm a dreamer, and even if they do it will probably have a giant fly in the ointment somehow. Still, the more people respond to forums like this, the better.




I think the message we'd all like to send is:

"Hey MS, You released Windows 7, great job. Now tons of people want to pay you good money under the vastly profitable SPLA program just to sell remote access to the product in VM (and MS office too).... awesome.... Please just take our money and make us legal."

Unfortunately, I think it's never this simple for a room of Microsoft roadmap and licensing planners. As Brian points out, their success has been based on domination and control of "desktop technology". VDI represents a rapid mutation of their business model. Their instincts are to maintain control through restrictive licensing. Their bigger goal is to slow and guide the mutation toward some model where they maintain domination. They could easliy make a lot of money right now off the SPLA of virtualized windows 7 if they gave us what we wanted,  so their long-term-master-plan is obviously more important than this comparatively short-term issue. To us, it's an annoying restriction, and they're playing big evil machine again. To them, they're just being careful. Thats me trying to remain objective. Meanwhile, I've already spent a few hours now ranting about how difficult this policy has made my life to my colleagues.




Your article is great, but I think that you are focusing on the wrong point. As Icelus stated "You are actually describing why Microsoft would hate Application Virtualization because it allows Windows apps to be run on non-windows client devices... Since VDI uses Windows in the datacenter, as long as Windows is used I doubt MS cares."

This is one of the reasons that I believe that the focus should be on Application Virtualization and delivery.

VDI, unless you set it up using non-persistant desktops, creates the same issues that the desktop environment has as far as infections,  troubleshooting, user profile issues etc... If you use non-persistant desktops you lose personalization, unless you add AppSense, and forensics abilities.

This is why I believe that Citrix and to some extent Quest is so far ahead of the game. They can deliver the appliactions to anyone, no matter what the OS. This is true delivery-as-a-service.