Desktop Virtualization is NOT about cost savings or saving money

A lot of people think that desktop virtualization and VDI are about saving money. But I don't think that's true.

A lot of people think that desktop virtualization and VDI are about saving money. But I don't think that's true. Desktop virtualization has a lot of advantages, but typically it ends up costing more money than traditional desktops. (But that's ok. It's fine to spend more money for more features.) But if you're trying to do desktop virtualization just to save money, you were probably fed a line of horse shit by someone. (Which is fine, as long as you knew it was happening.)

First, I should point out that I think most cost analyses are total junk whose results can be "gamed" to prove whatever point you'd like to prove. That said, a lot of people still try to do desktop virtualization to save money, (be it capex or opex), but desktop virtualization is not about saving money.

This is not to say that desktop virtualization is not worth doing. In fact there are a lot of advantages to desktop virtualization: Security. Availability. Faster installs, adds, moves, and changes. Easier disaster recovery. Flexibility. Reliability. But all these advantages (which are awesome!) cost money. And implementing desktop virtualization to get these advantages costs more money than not implementing desktop virtualization.

And you know what? That's ok.

It's totally ok to pay more for desktop virtualization since desktop virtualization gives you awesome new features. This is the way the world works. More features = more money. I often explain this in terms of cars. If you decide to replace an old shitty car with a new one with heated seats, bluetooth navigation, and computer-controlled automatic parallel parking, that new car will be more expensive than your old car. But that's ok because that new car has those awesome new features. So it's ok to spend more on the new car than the old car.

The same is true for desktop virtualization. Desktop virtualization gives you some of the awesome new features that I listed above. But in order to get those awesome new features, you have to pay for them. That's just the way the world works.

Now at this point you might be thinking that this is total hogwash--that "more money for more features" doesn't apply to computers what with Moore's Law and all. But that's hogwash too. Take Blackberries, for example. Giving your users Blackberries is more expensive than users without Blackberries. You have to buy the devices, the data plans, the Blackberry Enterprise Server, the licensing, the training, the support contracts… all of that is additional costs you have for Blackberries that doesn't exist if you don't use Blackberries. But does that mean that no one will use Blackberries since they cost more? Of course not! IT folks recognize that while having Blackberries costs more than not having them, they provide many new features that are awesome and worth the money. So we pay more for Blackberries.

So why isn't the same true for desktop virtualization? What crazy world do we live in where desktop virtualization has to be cheaper than traditional desktops? Why the hell do we want new features and higher availability and access from anywhere but then we STILL expect all that new capability to be cheaper?!?! What kind of crazy bizzaro world is that??

An attempt to save money with VDI

I was explaining this at one of our Desktop Virtualization 2011 events a few weeks ago, and someone challenged me on this. He said that for them, desktop virtualization was cheaper. He explained that right now they're spending $1,000 per desktop for new equipment every four years. If they go to VDI, they can buy $200 thin clients instead of $1,000 desktops. Then they can use some of the $800 saved per desktop to buy the back-end servers, storage, licenses, etc. for the complete environment. All in, they're saving $200 per desktop by going to VDI.

At first you might think, "ok, so for this guy, VDI is about saving money." But there are actually are several problems with this example and this "savings" is total BS.

First, you can buy PCs for $300 or $400. So I would argue that the best way for this guy to save money is to stop buying $1,000 PCs and to instead buy $400 PCs. Now you might argue that he actually needs $1,000 desktops--that his users are "power users" and that they actually need that much computer. Ok fine. But if that's the case, then there's no way in hell that $200 thin client powered by $500 of back-end VDI kit is going to come anywhere close to delivering the computing power and user experience they need.

So in this guy's case, he's saving money not by going to VDI, but rather by drastically cutting down the amount of desktop computing that he's delivering to his users. And he can do that with the $300 PC route--he doesn't need VDI to do that.

The REAL reason people "save" money with desktop virtualization

Ironically the only way people save money with desktop virtualization is because they use the desktop virtualization to deliver an inferior desktop product when compared to their existing physical desktops.

So sometimes people will say, "I'm going to VDI. That will be cheaper because we're going to have shared (and therefore locked-down) disk images which means the users can't screw things up, and that will be cheaper to manage than the current 'Wild West' world of personal images." Again, that's true. But the money savings is coming from the fact that you're locking down your desktop images, not the fact that you're using VDI. And again, if you just wanted to save money, I'd suggest that you lock down your desktops but that you skip desktop virtualization.

To be clear: there are a lot of wonderful and perfectly valid reasons to virtualize your desktops. It's just that saving money is not one of them. And if you use desktop virtualization as your excuse to completely overhaul the way that you deliver desktops, well it's the complete overhaul that's saving you money, not the actual desktop virtualization.

But at the end of the day, you gotta do what you gotta do. And if you really believe desktop virtualization is about saving money, then more power to you. I'll even help you prove it even though I think it's total crap.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

I guess it is hard to quantify the potential for increased productivity.


Hallelujah!  It's wake up and smell the coffee time!


Hopefully the DV hype curve will soon start to flatten out and people will start realising how complex and expensive this stuff is to assemble and manage.

I design and plug together this stuff all the time and I'm constantly dumbfounded that people really think that this is progress.

Forget saving money, I'd wager that over a 3 year period DV will turn out to be monumentally more expensive, Costs will also increase as the 'layer cake' model becomes reality and we push our data through a plethora of additional filter drivers.  We'll soon need a quad vCPU virtual desktop just to handle the overhead of working out where each block of data needs to go. Moores law will be essential just to keep us on an even playing field.



At least the DV hype curve pays the mortgage!




Your examples only scratch the surface. It's agreed that DV brings superior Management, Security, and Availability to your infrastructure at a cost. However, are you able to list another technology that provides the same functionality that DV does while having a lower cost?

I didn't think so.

The Blackberry example is a great one, and I will bet that when a solution comes out that offers the same or better features at a lower cost then it would make RIM obsolete. We are starting to see that already...

For businesses that require some or all of the superior Management, Security, and Availability that DV offers, they won't find a better bang for your buck because it's the only beast in town.

Sure, I can be DevOps and script my management, but I am disconnected when it comes to security and availability. It's about the synergy of the three.


Is VDI about saving money ? Probably not at first sight as you mentionned. BUT !...

We have years of experience on other desktop virtualization solutions (TS application virtualization, RDS published desktop and others) that are savings money ! Not me saying that... All customers i've talked to that have made the switch globally on it!

But inagree with you, the main benefits are not abkutnsaving money but more on doing thing differently with new capabilities you mentionned...


Lets say you have 1,000 desktops.  Instead of replacing them with thin clients after a typical 4 year refresh cycle you decide to convert the desktops to thin clients with VDI blaster for $30 you will save enough money to pay for the back-end infrastructure.  Then after those converted PCs begin to fail if you replace them with $200 thin clients you are still saving money over a $300 desktop because the infrastructure is already paid for and maintenance for the Virtualization solution would be less than $100 per year.


In some cases IT IS about saving money, but the ROI is very long. The TCO goes up, but eventually seems to flatten out.

The magic of HDV/VDI/DVI is that it enables backup or contingency plans for companies that cannot experience any downtime or disruption. This becomes both a CAPEX and OPEX distributed across multiple business units.  

Sure there are tools that can accomplish this in more cost-effective ways, but executives want to leverage their existing infrastructures and start dabbling first before they decide to go all-in. It is silly to think that the equation is a technology one and that is simply short-sighted.


Brian, first off,  an absolutely fantastic article on this topic. You hit the nail on the head. Here's my two cents(sense) on this topic.

1. People/Users don't give a sh*t about desktops. They care about applications.

2. VDI is another layer(cake layer) of unneccesary complexity in the environment.

i) most of the users in the world are mobile (or want to be). We see this with the explosion of tablets and laptops.

ii) What's a desktop folks? What do people do when they launch a desktop? Oh wait...they launch Outlook, Excel or some other application. Hmmmm...let me get this straight!  A user out there opens up his laptop(or ipad), logs into it, then opens up a VDI session to only then launch an application.

iii) Oh but my users can take it offline and work on a plane and then sync it back when they go back online. --OK, who cares? It's onerus for users to do this.  In the traditional way of laptops we can set their email/Outlook in cached mode with little to no user interaction.

VDI is merely another 'Published Application' that users use to launch (drum roll please) ...other published applications.

3 . with VDI don't forget about printers!  If you have 1000 VDI desktops then you better understand printers in your environment. I would rather manage my printers with a xenapp farm of 10-20 SERVERS than 1000 Virtual desktops. But then again I like to spend time with my kids and wife and dogs.

4. The Systems guy now becomes the helpdesk guy.

5. VDI doesn't address the endpoints. I give a user a corporate laptop with an image on it already so now my user logs into his personal laptop and then his VDI.  Ok wait a minute here! My team now has to manage 2 images for every one person...(the endpoint and then the VDI). -- Ok now don't tell me about one shared image, most users want non-persistent customizable images not a shared, always the same image.

6. Just because you've worked with these vendors before and they are offering a new 'VDI' solution or product doesn't mean you have to buy it.

7. VDI is a problem looking for a solution.

8. The Evolution of Technologies:  If one were to look at both solutions(VDI as it is now and traditional RDS/Xenapp) one would assume that VDI was introduced first in the evolution of technology. If the VDI model came out first ( before TS and Winframe) it would have been a great product.  If RDS or Presentation Server(Xenapp) came out after VDI it would have totally killed the VDI model as it is currently . THE RDS and XENAPP model is the smarter model that has been around for a longtime and is PROVEN actually should have come out after VDI.

9) VDI = Very Dumb Idea



I enjoyed your "Desktop Virtualization 2011" event yesterday on 6th Oct here in Frankfurt. Made up my mind being a little more critical on VDI.


But I'm missing something in this discussion concerning the costs of VDI: In my opinion there is a difference between investing costs and operating costs. The IDC pointed out, that IT security, high availability and lower operating costs are the main reasons of german enterprises for VDI. So ok, I agree that higher security and availability are better features, so higher investing costs are just fair. And in the long history of electronic technology we should be used to high prices for new technology, although the production costs of that stuff possibly are absurdly low (I'm thinking of thin clients).

BUT what about the operating costs? What about the costs for let's say 5 - 10 years? Let's see, what perhaps make sense:

a) power costs:

I believe it, if people say that dynamically hardware ressources (load balancing) and thin clients is a good solution against unused hardware ressources ( = power saving ). But I also think the VDI software will cause enough load to equalize this power savings.

b) personnel costs:

That's not easy to estimate and of course every enterprise has individual needs for client administration and troubleshooting. But fixing a ruined virtual desktop by switching to another virtual desktop sounds pretty good to me, as well as a central VM image where updates, configurations and stuff have to be done just once. On the other side, as Brian pointed out VDI is a complex thing. Not only complex to build up, but also complicated to understand and to administrate.

So, concering personnel costs I'm not able to express a confident final opinion. Maybe you can?

c) hardware (server & client):

What SEEMS VDI to be?

First you build up the VDI server (maybe some terminal services too) and a central storage solution. So this would be the same as buying a new desktop PC generation, cause I buy new hardware for my desktops even if it's server hardware in this case. On the client side I could use my old fat clients furthermore (without the unnecessary hardware components) or I could switch to thin clients or zero clients.

Of course the total investing costs (VDI software, licences, hardware, . . . ) would be quite more expensive than a new PC generation.


-> If my virtual desktops have to be more powerful after let's say 5 years because some new application-versions need more ressources, then I have to expand the servers. That would be cheaper than a new PC generation.

-> If I need more storage, then I just have to expand the centralized storage servers.

-> I don't have to buy new clients for a long time, cause they just handle I/O. And I think there are rare and unimportant changes in I/O technology (Do I need USB 3.0 as soon as possible?)

So hardware becomes more modular. It would be devided in cpu + memory, storage, I/O clients. If one of them is getting weak or old, I just have to expand or replace this kind of hardware and not all of them.

I pointed out that it just SEEMS to be like this. Since the event with Brian I'm not quite sure.

For example I have some doubts concerning one of the most important cost saving point: hardware. Especially on the client side. I read that some (or maybe all) zero clients are related to concrete protocols or even VDI producers, like VMware. Some manufacturers of zero clients express like "Works fine with . . . ". So I would tend more to the thin clients. But I see also a general problem concerning VDI and clients. If I switch to VDI, someday there is no other sensefull way for me than using thin clients, I believe. And then I have to trust the vendors of thin clients and VDI solutions, that I'm not forced to buy a new thin client generation atfer 2 years.

Has somebody an idea why or how the vision of an 10-years-old thin client, which is still up to date, could be just a fantastic dream?

Cause in this case I'd like to be woken up.  ; )


There is ***loads of research available on the TCO of HVD\VDI  implementations compared to regular desktop TCO and they all seem to back Brians statements.

Only when you consider well managed, pooled virtual desktops with state of the art image management software, will you be able to reach TCO numbers that resemble desktop TCO. So you dont have to ask yourself if it is about saving money. It just isn't.'

In addition keep in mind that you will be adding additional management complexity to you desktop environment because you will end up with a mixed environment (no 100% use case for VDI today). Even VMware isn't able to eat its own dogfood to its full extend.

So indeed Brian, you need to think twice before you move to VDI and by thinking twice you will maybe make the right  decisions regarding  your desktop management solutons and approach. This will in the end result in the needed savings. Not VDI itself.


Totally agree that VDI is not about saving money - because just to start with, the infrastructure costs are going to hammer your capex budget with all the server, network, and storage investments.  It's hard to show a return from such a big up front investment.

IDV or Intelligent Desktop Virtualization is a different animal though.  Because it gets rid of the up-front infrastructure investment, but still delivers all of the management benefits, it really delivers ROI pretty fast.   Just image VDI without having to pay for the servers, and you can see what I mean.

With IDV, the desktops are still virtualized and the images are still managed centrally, but they are just executed locally on the endpoints.

Brian's got a great point that PCs can cost as little as $300-$400 these days.  With Intelligent Desktop Virtualization, these PCs are effectively turned into managed devices where you can execute your virtual desktops.  And by moving the desktop execution locally - voila - you save on the infrastructure, which generates the ROI.   And an extra benefit... you get better performance because the desktops are running natively - it feels local because it is.


@Sham sao

I fully agree. Leverage your existing desktop infrastructure with centrally managed and locally executed layers. Best of both worlds.


Hi Brian,

I am not sure if this is the correct place to post this question, but i could not find a better blog to do it.

We are planning to migrate our company to Microsoft VDI environment as we already have the license for it. But before doing that i wanted to know the tools which can utilised o assess my environment if it is cabble enough to move to VDI.

Could u give some inputs on this..




Wow, VDI sure is taking a beating this month — see also Simon Crosby's "VDwhy?" post and my "VDI: Virtually Dead Idea" followup:

I agree that VDI simply adds another cost, but I'd also like to take a look at your purported benefits:

Security: False. If anything I'd say that compared to (multi-user) Terminal Services the security is *worse*, even if only because there is less emphasis on security when you have dedicated machines you can blow away.

Availability: False. Many of the things that will take out a physical desktop will also take out a virtual desktop, only now you've got an additional layer of virtualisation — complexity is the enemy of security and availability.

Faster installs, adds, moves, and changes: False. You still need to provision a physical device and the difference in effort between deploying a thin terminal and a PC (with sensible imaging infrastructure) is marginal.

Easier disaster recovery: False. Furthermore, disasters now affect many users, while thick client outages only affect a single user.

Flexibility: False. If you have a sensible imaging system you have just as much flexibility as you would with virtualisation, only without the additional cost and overhead.

Reliability: False. Desktops are relatively reliable, and often survive their entire life with a single image. Problems only affect the local user and are trivially resolved. VDI problems on the other hand (including performance problems which are notoriously difficult to diagnose and resolve) tend to affect everyone.

In summary, many of the arguments put forward for VDI today are, in my opinion, bogus. Compared to mutli-user Terminal Server installations I see VDI as a regression, except perhaps for end-user facing services where a high level of isolation and configurability are demanded.


Sam is spot on people.  Read through what we wrote carefully.  It's all true.


While I agree with some of the comments, and all may be true in certain environments, there are other aspects you're missing.

Security - Yep, I agree with this whole heartedly. There is nothing more inherently secure about VDI vs. a normal desktop. The reality is if your users have admin rights today, VDI may be the excuse you need to lock those desktops down. Plus the features typically implemented with VDI are the tools you need to make locked down desktops work (profile and application management).

Availability - I disagree, as long as you have the capacity you MTTR is going to be much faster for a given failure.

Faster - depends. I disagree in certain environments. Consider a large organization, maybe even unionized where it may take weeks to get a desktop delivered, hooked up, and added to the domain. Much less the time and cost to reclaim the old device. Yes you still have some of these issues on first time delivery of a thin client but their  lease lengths are typically longer and do not require a remove and add for a new user.

Easier DR - it depends on how you look at this one. The aspect you mention of affecting many users at once is a huge problem and must be designed around in the implementation. (As some vendors out of the box totally miss this aspect!)

on the flip side if I have a hurricane that takes out a complete call center I can easily bring up another location for those relocated or other workers to take on that workload.

Flexibility - ?? What imaging system could possibly give you the flexibility of VDI? Altiris/PVS type solution is the only thing that could even come close. This and MTTI are the saving grace of VDI and why, if you need it, it makes sense to make the investment.

Reliability - I agree here and goes back to the availability statement.

Summary - I agree! The "VDI is a solution looking for a problem" fits here. User density is much better in XEN/RDS environment. You just have to do your due dilligence here, run the numbers, and see what works for you. Honestly, XenApp/RDS licensing is so expensive that the cost (if you still do not need that XA license!) are close to a wash.


it is very important thank you!!

see this:


thnak you it is very important