VDI is 11% more expensive than traditional PCs. Why? Increased costs of Microsoft licensing!

Yesterday I learned about a study that Microsoft released (PDF) that compared the TCO of VDI versus traditional desktops. Their analysis was that with Windows XP SP3, VDI was 9% more expensive overall, and with Windows 7 VDI was 11% more expensive.

Yesterday I learned about a study that Microsoft released (PDF) that compared the TCO of VDI versus traditional desktops. Their analysis was that with Windows XP SP3, VDI was 9% more expensive overall, and with Windows 7 VDI was 11% more expensive. Now of course it's well known that I believe it's easy to manipulate TCO numbers to show whatever you want (see How to lie with cost models, Dec 2009), so what's really interesting about this study is not that VDI is higher, but that it validates that Microsoft prefers a world of traditional desktops instead of VDI. (Although was anyone really questioning whether that was true? I mean if Microsoft really wanted VDI to take off, would VECD, VDA, and MDOP exist?)

Also interesting was that Microsoft's study confirmed that the money saved on thin clients was more than enough to offset the additional server, storage and networking costs. They also confirmed that VDI environments would have lower desktop support costs. So how did they end up with VDI being more expensive overall? Microsoft blamed the "additional licensing costs associated with virtualization" as well as the "more demanding desktop engineering requirements due to the complexity of designing and managing the VDI environment."

Quantifying these numbers, the Microsoft study showed that "VDI reduces hardware costs by 32% but increases software costs by 64% canceling any savings." At this point it's probably worth mentioning that study reported the additional licensing costs as "Additional licensing costs associated with virtualization, management and desktop software from VMware and Microsoft®." In other words, VDI is more expensive because Microsoft is making it so! (Well, at least that's a contributing factor.)

Of course it's easy to poke fun at this kind of justification, and it certainly makes for a catchy headline, but of course Microsoft software licensing isn't the only cost increase with VDI. The Microsoft study also talks about the diminished user experience with VDI when compared to a well-managed PC. (And those of us who've been doing this for awhile know that's the case -- our "best case" for remoting protocols is that they deliver a user experiences that's just as good as a locally-running desktop. :)

In all seriousness, this Microsoft study is really, really good. It outlines a lot of the harsh realities of VDI today based on polling of 500 organizations in the US with at least 500 VDI users each, and of them the study mentioned that, "Their opinions are based on perceptions of their currently deployed environment and not on benchmarking." This study certainly reflects many of the harsh realities of VDI today, including that while it sounds great in theory, for most people VDI is complex, expensive, and delivers a worse user experience than traditional desktops.

So definitely take a glance at it. The whole study is 36 pages, but it's worth a quick skim. What do you think about it? Does the author bring up valid points, or is Microsoft just trying to defend their traditional desktop dominance?


Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Some of the points have merit but some quick observations to point out...

Surely "office workers" would mainly be on RDS and only use VDI when needed? Are people really trying to take the VMware approach to desktop virtualization?

Again in the "best practices" section, pushing SAN storage... We're seeing rapid cost savings using local/solid state.

I only had a quick glance so will have a proper read later.

Well written but with questionable content!

MS is a divided company... On the one side you have a section pushing VDI/RDS and promoting the nearing RemoteFX stack which is some ways is MS's stamp of approval on desktop virtualization. On the other side you have the guys that wrote this!


I've studied this document in some depth and it's quite clear that anyone with even a basic understanding of desktop virtualization can poke holes in it pretty much anywhere they want to. Take that starting point and the question is then why is it so wrong.  Conspiracy, incompetence, naivety?  

I wouldn't presume to suggest which one of those it might be, but given the fact that it is a Microsoft document it's highly likely to be used by those with a vested interest in maintaining the status quo of non-virtualized desktops as a way to e-mail proposals to explore desktop virtualization.





Ah, but a great conversation starter!  Great find, Brian.

Inherently, we probably all think that VDI should reduce IT manpower through centralization and improved standardization.  And yet here they are showing in anticipated increase in IT Operations (page 11).  It isn't until I look at the detail (page 16) that this begins to make any sense to me.

In the breakdown of IT manpower, while the need for service desk workers is reduced, the demand on desktop engineering is increased.  The detail on how this happened was captured in Appendix C, where I see a major increase in Tier 3 support.  So while I'm not sure about the numbers and conclusions, the idea that IT operations costs might not decrease as I expect but increase simply because higher paid people are now dealing with the problems is one to consider.  

These may be more short term horizon numbers for IT than long term.  Moving to VDI will mean that IT operational structures will have to evolve to become more cost effective over time.  What a great discussion opener!


Let's keep in mind though that VDI is more expensive overall, none of our VDI projects have been driven by lower cost, they are always other factors

In the end you are ADDING infrastructure to do VDI and you still have an endpoint device

VDI is another tool in the kit that addresses use cases that didn't have a tool before, in my slice of the world it is something like 10% of user base


@ Steve

Even if you accept Microsoft's position that desktop virtualization is more expensive than traditional solutions; the use case that it supports must deliver a (measurable) benefit to the business or there is no point in adopting it. of course it's up to you whether you want to go to the trouble of quantifying that benefit, or just go with the gut instinct that it delivers something that business must have regardless of the cost of, ultimately it comes down to the bottom line of the business and not just the IT department's budget.


I've done several VDI ROI analyses, and there are certainly many situations where VDI saves both CapEx and OpEx (and others where it doesn't). One of the key factors is whether or not the organization is already using a highly managed/locked down physical desktop environment. If not, the VDI ROI benefits from all of the cost advantages from the management capabilities it inherently entails. The ROI also gets a boost when VDI is deployed as a component of an overall virtualized strategy rather than as a silo’d project. Deploying virtual desktops as part of a virtualized data center leverages (at least to some extent) existing compute, storage, licensing, management tools and IT staff expertise.


I'm with @ Steve

When the use case applies to ~10%, the ROI and benefit measuring studies themselves will erode any savings quicker that you can say RDS. Rely on the gut feel of people who are experienced deploying the technology....not those who do ROI studies !!

Personally having VDI as an option has allowed many more "yes we can" conversations.


I've read this document two months ago and was suprised about the result. And since one week I am even more suprised because I've tried to understand VDA, Software Assurance etc. and it showed clearly that Microsoft does everything to push VDI because of their licensing modell of VDA and SA. We don't have SA and therefore, we had to buy a lot of VDA licenses just to use VDI. Or we get SA for all desktops and do have a lot more benefits (MDOP, App-v, etc) and it is way cheaper!

I've tried to create my own ROI calculator for our company to check some scenarios but it is just impossible to be neutral. The most difficult part is the calculating of support complexity. Is it increasing or decreasing without VDI, only VDI, mixed environemnt, little bit RDS, only RDS etc.? How can I compare a VDI administrator, RSA engineer, help desk, etc. in terms of cost savings?

At least, this document is an interesting approach to include the differences in support complexity. Until now, it is the only study I know which give it so much thought.

At the end of the day, it is still unclear what is the best client strategy for my company... However, it is always great to read such discussions on this board and helps me to get a better feeling :-)


It amazes me that MS still does not get that this is not just about VDI. They have RDS and could build a compelling story that makes it far cheaper than a PC. It's also not just about cost, there are a ton of other benefits that are worth paying for. Too many naive people in this industry and that's the sad reality of what the average person believes when they see such junk from MS.


1) It should be noted that this study was done before the notorious July1 announcement about SA virtualization rights and VDA. Depending upon the customer licensing situation, the software costs for VDI could be much larger.

2) Typically, VDI studies only use the KWh consumption costs for PCs and VDI Infrastructure. This DRASTICALLY underestimates the cost of running VDI infrastructure in a DC. As a rule of thumb, 1000Kw costs at least $10,000 of capital to provision in a DC.

3) It is doubtful that the VDI deployments condsidered that Backup VDI infrastructure must be deployed for disaster recovery.

4) VDI can almost never be a total replacement for conventional client methods and the study rightfully describes that the environment becomes more complex and requires more technical skills. It is a supplemental architecture that can be justified if there are EXCEPTIONAL business requirements such as security, offshoring or mobility (this is big).

5) The study says that the network must be enhanced. Incrased bandwidth and availability of the network is a requirement with or without VDI.

Clients are dead in the water if they do not have network connectivity.

6) For any business with conventional client access, VDI is not going to save them money. Furthermore, they will actually lose money if they are already "Rationalized".

7) It's a bit naive to use MSRP for all of the components, but it helped reduce the complexity of this analysis.

8) It's worth reading the Appendix E for the contributors' comments.

9) Consider MSFT's onerous pricing for VDA of $100 per year (e.g. $700 for the 7-year life of a thin client and $300 for the 3-year life of a smartphone). You can be sure those costs were not included by the surveyed customers.

After VERY carefully examining the EULAs for Windows 7, I now believe that RDS into non-virtualized clients is probably the appropriate strategic direction for users with critical business requirements such as security, offshoring and/or mobility. BTW, if a PC (or blade) is going to be remotely accessed by multiple users (ignoring service desk), then an OEM license for Windows CANNOT be installed on that PC (or blade). Profile management appears to be necessary in order to rapidly recover from a hardware failure.

Are we having fun yet?


Virtualization doesn't have to be an all or nothing proposition. It may be appropriate for some applications, and not for others. This article suggests that SMBs, in particular, can benefit from software savings by analyzing the needs of specific users and virtualizing where it makes sense..




If you read IDC "Quantifying the Bus iness Value of VMware View" published on Sept. 2009 you'll find out that VDI is cheapper.

At the end every organization should make his own study / analysys.