Results of our Citrix XenDesktop 4 Licensing Survey. The community has spoken!

The results from our XenDesktop 4 Licensing Survey are in with over 280 participants. As expected they were largely in favor of Per User licensing.

The results from our XenDesktop 4 Licensing Survey are in with over 280 participants. As expected they were largely in favor of Per User licensing.

Haha...little blogger humor.

In fact, the results were overwhelmingly against per user licensing, even more than I thought they would be.  There were are few people that said the licensing model works better for them, but it was dwarfed by the number of users that said it would not work at all.

Let's take a look at some of the questions. If you'd like to see the entire survey, download the attached PDF which contains the results and text responses to questions 4 and 9 (they're at the bottom).

Question 3: What is the best description of the use case for which you first deployed (or plan to first deploy) virtual desktops?


This echoes what we saw in the comments for Question 4 as well, except that training and disaster recovery also appear to be high on the list of use cases. Not much telling here, other than confirming the use cases that we all suspected.

Question 5: Do your virtual desktop users share the same endpoint devices (ie for lab use or shift-work)?


Here's where we start to see why people are resisting the Per-User license so much. 63% of organizations have users that share endpoints. Now, I'm sure not all of these are shift-based, in fact there's probably a good many of them whose users sit down at whatever endpoint is available. That means the results here might not totally indicate the people would like to see a CCU model, but they might also be warm to a per-device model.

Question 6: If you are using or plan to use VM-based virtual desktops, how do you intend to deliver most of your VM desktops?


Now we can start to see what all the hubbub is about. Most XenDesktop users expect to employ non-persistent/pooled desktops in the organization, rather than persistent. What this boils down to is that they expect to use XenDesktop in much the same way they use XenApp, while taking advantage of the XenDesktop feature-set.

Question 7: How would you view licensing a desktop virtualization solution on a PER-USER basis? That is, a license is required for every user that accesses a virtual desktop but that user can use any endpoint in any location.


This chart needs some clarification before we go into it. Participants rated their view of Per-User licensing on a scale from from Very Unfavorably (1) to Very Favorably (5). As you can see, a full 50% (140 participants) rated their view of Per User licensing as Very Unfavorable, while only 8% (22 participants) said it was Very Favorable.  In all, 72% of respondents were not in favor of Per-User licensing, while 20% were. The other 8% are Swiss.

Question 8: How would you view licensing a desktop virtualization solution on a PER-DEVICE basis? That is, a license is required for every physical device that accesses a virtual desktop but any user can use that licensed device.


The same clarification applies here. Participants rated their view of Per-User licensing on a scale from from Very Unfavorably (1) to Very Favorably (5). While a Per-Device method is more agreeable than a Per-User method, the numbers still suggest that people don't like it, either. 50% of respondents (140 people) responded either Very Unfavorably or Unfavorably, while 32% (90 people) said that a Per-Device model would be Favorable or Very Favorable.  18% said they were Somewhat Favorable, which means they're also somewhat unfavorable in this unscientific survey.

Essay Questions

The essay questions provided four or five pages of entertaining reading, with a few colorful remarks about Citrix and the products, but there was quite a bit of useful information in there on use cases and how the change to Per-User licensing affects people's current and future XenDesktop deployments.  Not wanting to take an unnecessarily negative tone, let's look at the positive responses first (I think these are the only 6, and we had 96 responses to question 9):

  • "It's actually fine for us because for budgeting purposes it's easier for us to account for a license per user than it is for some large arbitrary pool of licenses that will 'cover us'. It helps to know exactly how many users we have licensed for auditing purposes."
  • "For me a per device case makes more sense as XD will be used to REPLACE desktops so their idea for licensing makes sense. It's only a [pain in the neck] if you're buying XD with the main idea being to use XenApp for your applications. If you're did this you're dumb though."
  • "because 60-70% of all users are logged in concurrently and we'll get a 2:1 trade that allows more flexibility, the move is good for us"
  • "The per user (named user) licensing model would allow us to assign licenses directly to users when odering our VDI product internally. That's good for us. The price for the licenses will put us off evaluating the product"
  • "Named user licensing is fine - they should also offer named device licensing as well which would allocate a license to a named connecting device"
  • "Per user pricing makes the most sense here due to eventual likely deployment to mobile devices (both laptop and smartphone)."

Clearly, for some users, this makes perfect sense. Companies with many users using multiple devices, a high proportion of virtual desktop users vs regular users in their company, or those with big-time budgeting headaches would much prefer a Per-User (or even Per-Device) model over a CCU model.

Now, let's take a look at a small portion of the negative responses:

  • "Depending on how the licensing shakes out by the product release date, it could have a large impact. Politically we are still fighting the battle of VM View vs. Citrix Xendesktop. This could make the decision sway toward VM View (which I would support even though I find XD technically superior) VDI is already a tough sale; last thing we need right now is have to deal with this."

(this was said over and over - that people will stop and take another look at VMware View or scrap the XD plans and move to View. Some even said they would switch even if the solution they found wasn't as feature-rich as XenDesktop.  Have fun with that one, AppDetective.)

  • "per Concurrent user is better than device, Named user doesn't fit our current use case."
  • "With Citrix going with PNU licensing on XenDesktop, it doesn't make sense anymore to migrate from traditional desktops to virtual desktops. The savings is not there anymore. It'll cost more to go to VDI, so we'll just have to stick with Traditional Desktops."
  • "Citrix has repeatedly costs more and more and due to this we are trying to get away from using Citrix. Before we were forced to use it because of it's functionality and the features it had. But now with 2008 Server, we can offer everything Citrix did for us without the added costs. We do loose some features and administrative logging, but for the costs that we save it is well worth it."
  • "Per User at the current proposed price is overpriced to be adopted widely. Additionally, it falls flat in Citrix's major verticals of Healthcare and Education, where high concurrency is a fact of life. ... And changing the pricing model during the last quarter of the year forces everyone who had XenDesktop in the pipeline for an end-of-year budget dump purchase to rework their cost model - with less than 90 days to go to purchase.  As a reseller, we'll spend more time re-educating the customer on what they were already sold on than we will actually closing the deal."
  • "We want to provide a remote access VDI solution to 3500 users, but we only ever have approximately 100 users online concurrently. Instead of buying 150 CCU licenses, I now need to buy 3500. In this current economic climate, the licensing change would blow my budget to the point where the project will be scrapped."
  • "My company deals with time zones, and the concurrency license aids this greatly - As east coast users log off, the bulk of the west coast just ramps up - the per user model will most likely prevent adoption of the technology"
  • "The only licensing method I would consider is per-concurrent user. Period."

Still, other users were receptive to change, just not as drastic.  Some asked for Per-User or Per-Device so they could decide how to best implement it. Other suggested a Per-VM license, which I suspect works out to being very similar to CCU, but does offer up another price model and other options.

So there you have it.  For the full report on the survey results, download the attached PDF file. We have the ability to scrub these results by the answers to certain questions, but it's a lot of work to do it for all of them. If anyone cares to see, for instance, how many respondents in the Education industry said that they favor Per User licensing, let us know in the comments and we'll try to make that happen.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Good work Gabe! Clearly there is a clear and valid message here to Citrix. You need a CCU option if you want to remain relevant. Citrix line up all your decision makers tomorrow have them READ this, kick them in the balls several times and ask them to do the math, and every time a marketing statement comes out of their mouth kick them in the balls again. Repeat this process until they get what happens in the real world. When this process is complete, tell them well done for the per user option it's good ONLY when you are gong all our VDI, kick them in the balls again and explain the value of options for people who are trying to use your stuff to get there and kick them in the balls again for the hell of it!

I find the results of question 6 very interesting. It tells me, people don't understand VDI as it stands today. I've said it many times before, pooled desktops at scale are just not possible since most people have no idea how to drive such a process with so much variance. I bet these people are all XenApp mind set types that are just trying to create a XD option. Pooled is a future state, not the norm today, unless you are very small or simple, which may be the case to be fair to those who posted.

Drive to VMWare, Why? I would expect people to go MS/Quest/Ericom etc, VMware screws you in other ways. Cost of ESX, lock in and on on on. PCoIP sucks balls. There's a difference for settling for less features and completely sitting on knife for fun.......


The drive to VMware or Microsoft is driven by reducing the vendors in your environment I suspect, and VMware was predominant because View is generally better known than Ericom, Quest etc.  People are comfortable with VMware and have support contracts with them.

In relation to question 6, I answered pooled desktop however for pooled to work you need to completely assess your desktop ecosystem which is what I am doing in a couple of accounts, so I am not taking a XenApp mind to it, I am aware of what needs to done :)  However I have seen allot of XenDesktop implementations where they are not using pooled as the change was just too great and the organisation could not adapt to it yet.  But the future state was going to be a pooled environment for them.


It also seems weird to go in that direction - the VDI og SBC supplies the tools to move as many licenses as possible to the CCU-licensing-model.

So how on earth citrix found out that they want to move the opposite direction is a complete mystery to me.


What I find interesting in some of the comments (for Q9) is that people say that they cannot use per device model (or per user) but have to have CCU.

Which is otherwise fine, but how you then deal with the actual OS licensing as VECD license - absolutely needed unless running server OS as desktop - is per (client) device unless I have mistaken!? So you are forced to take account of  those devices anyways!

Or is it like I suspect: people either don't understand the Microsoft licensing involved or don't care (in which case they are setting themselves up for a big trouble down the line, come audit time..).


Kalle - In my opinion there is no reason for others to follow the footsteps of Microsoft. The fact that one SW-vendor makes it expensive to have VDI is no reason for me to ask other vendors to do the same. And thereby making my Cost-Benefit on VDI more difficult.


I want to update everyone on the status of Citrix's version of this story. Since I wrote the original article, Citrix has decided that they will publish their own results too.

They're claiming that they never "confirmed" that they wouldn't (like I had written), but instead that they just weren't sure and then after we talked to them that they decided they would publish the results.

I don't want to get into a "he said/she said" thing with them. Suffice to say they think that I acted irresponsibly in my post claiming they said they wouldn't, and I think they're just trying to save face by putting a spin on it after I made a big deal out of it.

Either way, I hope they do end up publishing their results and I'm curious to see how they compare to ours.


I was seriously concidering deploying Windows 7 XD 4 private VDIs as an option to XA 5. For us as host in the verticals the PNU or PU is nothing else than business as usual with the whole SPLA thing.

Now that the community,and hopefully a lot of "real" people ;-) are on the uproar I have halted my plans.

I simply cannot go on betting on the horse when rumours are that it has the flu and might even be dead for the time of the races.

Now I hear that there might eventully be a vaccine, but is it tested and will it arrive on time? Only time will tell, and the end is nigh :-)

In other topics. Could Brian/Gabe please give AppDetective a column on your site called "AppDetective's Rants" or something more catchy? I'm serious! No blog post is complete without the smart and funny wit of AppDetective.




If Citrix releases results that show users are in favor of the new model, I wouldn't be suprised.  Otherwise, how could a company make public a document that shows how pissed off customers are and not fix the issue?  Right now, the only stupid thing VMWARE could do is match Citrix's license model. Surely they want to move in for the kill.


I agree with everything AppDetective said (sans all the balls kicking - might want to get some counseling for that -  seriously something wrong there LOL)




Of course not, but if you intend to run Windows workstation OS in your VDI VMs, you still have to take Microsoft's licensing into acount. Broker's licensing model has no impact to this whatsoever.

Just to make it clear, I'm not big fan of VECD (and other "hidden" licensing costs of VDI) but I just wanted to point out that having CCU licensing on broker does not mean you don't have to deal with per-device licensing somewhere else. So those that have unknown number of client devices are in trouble anyways.


Come on guys just think about XenDesktop as a strategic platform and not some niche 5-10% coverage of an organzation's users.

We have customer who is planning on deploying XenApp Platinum for all their 1000 users. We told them that now they can get XD 4.0 for half the price and he could not believe his ears.

Licensing costs dropped from $600,000 to $350,000 (list price)!

So is Citrix trying to maximize their income or are they really trying to give customers a cost effective solution?

You should seriously look at how you have deployed Citrix at your site or your customers. Half of the blame is on you.



I largely agree with your commentary with the exception of one thing:

"I've said it many times before, pooled desktops at scale are just not possible since most people have no idea how to drive such a process with so much variance. I bet these people are all XenApp mind set types that are just trying to create a XD option. Pooled is a future state, not the norm today, unless you are very small or simple, which may be the case to be fair to those who posted."

This 'large variance' does not apply to a well run organization in the creation of pools. If you are doing internal VDI pools with known endpoint devices (i.e. a standard device schema like a Dell D620 lappy386 across the board) you have very little problem with 'variance'. I would venture to guess that the work that goes into image building (and locking and regression testing) that the current homogeneity of your network wouldn't be a concern. This is the only place you would create pools right? External serving and 'virtualization' is the XenApp, RDP, GotoMyX realm of screen scraper tech.

Caution Rant Ahead: I have a problem with most VDI and resource pooling. If the VDI solution I am using doesn't allow me to install over a large footprint of device types, what use is it? Are you telling me that you have to have homogeneous hardware to run VDI? I ask you then, what's the point if I then have to start tracking device types and builds? Why would I even need VDI instead of just a good packaging solution like WPS (Wise Package Studio) or MS Installshield?

Meh, whocares? This whole thing is moving too fast for me anyway. I am also getting tired of the re-branding being done by Citrix. I am very confused with their products now and I can imagine that management types are not very different. They are going to have to take me out for a few more beers this time around to start selling this crap.


Please allow me to add some sanity here. I do question the result for the following reasons. This is a classic case of using stats to mislead:

1. This is by no mean a scientific poll. In fact, majority of Citrix customers don't even know this blog. It captures a certain group of users who do not have a wider view of business.

2. Those who are upset will vote more often than those who are happy. The result is 100% certain to be skewed.

3. You have competitors who would love to screw with the poll to mess with Citrix.

It's very clear that per user (NOT per NAMED user) is the right way to go. Citrix does not want to play in the terminal services game any longer, it wants to move itself to be a key supplier for desktop solution.

Also, if you want to be seen as credible, you need to disclose that you have a certain captive audience and the poll is not scientific. If not, you are only misleading people. Take some classes on polls and statistics, those will do you good. I know for a fact that University of Iowa and Iowa State University both have top programs on how not to be fooled by these numbers.



-who do not have a wider view of business.

Really?!  Your here on this website, are somehow more informed than your peers?  Please do not attempt to speak on my behalf or others.

#2 and #3 yes, your points are valid and it is not scientific.  That said, it does not mean that all the responses, or any, were fudged.

The Licensing model is "PER NAMED USER" i.e. every AD account that is going to access the environment.  Don't play on semantics or suggest that a long time independent player in this field is fudging the facts.  

It was stated that this was an unscientific poll.  

It was actually a brilliant stroke of genius to do this poll as it DOES put Citrix in an awkward position with their results.  Citrix is not known to own up to marketing blunders and a poll that shows overall rejection of this is not likely to be disclosed.  

What Citrix wants to do with TS is not relevant to me, or others, who are not happy with the recent changes.  We want the original CCU model, or the option, returned so that our VDI POC's are not blown out of the water.  


@Rebecca:: you cannot stem the tide.. the community has spoken. We are the insiders of the app delivery world. Give us the CCU option back.



Nice work on the poll.  I look forwards to seeing if the different demographics between Citrix poll voters and your poll voters does anything to skew the results.  It's a pity that Citrix didn't ask for more demographics to help us understand how this affects the market in terms of number of possible licenses sold/not sold.

My own perspective is that, if you take into account the fact that the vast majority of XenDesktop sales are to existing XenApp customers, the number of people who are negatively impacted right now is likely to be rather smaller than we might expect from reading the number of comments seen here and elsewhere. Long term the picture could be different especially if Citrix's competitors retain their concurrent licensing model, but that is be no means a given especially when you consider the number of delivery models under the VDI umbrella .

Chris Wolf makes a strong case why customers should require per user licensing on his blog and I show why vendors will more than likely be required to move away from CCUs on mine. Chris's comments are well worth reading, I'll withhold comment on my own blog.

Regards Simon


@ Mike Jennings & Rebecca.  Licensing is:

If multiple users are sharing the same user ID, they each require a XenDesktop 4 user license. If multiple users use the system at different times, as is the case with the concurrent desktop model, they will also each require a XenDesktop 4 license. If a user connects to their desktop(s) and/or apps with multiple devices (e.g. desktop PC, laptop, netbook, smartphone and/or thin client) they need just one XenDesktop 4 license.

Its not per Active Directory user, it is per PHYSICAL user.

Refer to : under the heading Footnote: How exactly does Citrix define a “user” for XenDesktop 4.



Wanted to let you know that I did, in fact, say that this is unscientific when I said:

"50% of respondents (140 people) responded either Very Unfavorably or Unfavorably, while 32% (90 people) said that a Per-Device model would be Favorable or Very Favorable.  18% said they were Somewhat Favorable, which means they're also somewhat unfavorable in this UNSCIENTIFIC survey."

Not that we really needed that--not many scientific surveys lead with a joke about the survey's results or call out the neutrality of the Swiss.

For the record, we did take some measure to prevent stacking the deck: the results are deduped based on IP address to prevent one organization from stacking the deck from one location.  Not perfect, but better than nothing.  


@Gabe @Mike Jennings

Sorry I miss the note about this poll being unscientific.

You did get a ton of mileage out of this and I applaud you for driving a ton of attention.

By the way, what business are you in?? Why are you so mad??

It's just like whenever there's change. People who are comfortable with the status quo will always yell.

My company already uses XA Platinum on a one-to-one basis. Many of my peers in fact are doing exactly that. Citrix in my opinion is just adhering to what most companies in the world do. As a result, we get two licenses for one!

I cannot find any large scale companies today still use CCU. It's a model that will go the dinosaur age.  

If you are in a business role, you are thinking about how you can be more effective when it comes to managing your existing PCs. That's where Citrix is moving towards. Not just trying to solve the edge cases.

Remember, even when they claim they're in 99% of the Fortune 500, only a small group of users within those companies use Citrix. In order to move up the food chain, CHANGE is needed.


A couple of things. There is a lot of confusion on what Per User means. LET IT BE CLEAR, this is NOT a named user license which requires you to maintain a list. So for example if you have 10,000 people in your firm and want XD for only 2000 of them and you know who they are. Then you need to buy 2000 licenses for those folks. So you ask isn't that Named user? NO, the difference is if any of the other users want to use the 2000 licenses, you can swap out that unique user. Now you will ask isn't that like CCU. I will say sort of, but without the freedom to do what you want. How they control, honor system from what I can see is not 100% clear to me yet. So if Citrix can help clarify how the swap in swap out process or control works that will be great.

To all those that support per-user, I agree with you. It's great that Citrix are doing it. However people need a CCU option and I think per-device also makes sense. It depends on your use case. I will say to Citrix, Flex Cast is great if the licensing also matched the flexible delivery options. So Citrix needs to enable Flex Cast adoption, and cease the opportunity. Don't give this gift to VMWare. I also think CCU with XA can cost more than regular XD with CCU to justify if needed. Some will hate that, but at scale discounts come in, per-user ultimately makes more sense, and it fits with what they already have with XA. Flame away please, I would love to see what people think.

@James I think the piece you are missing is that most organizations have sunk cost into existing systems management tools and process. Assuming that VDI will come along and replace all those, and allow for variance etc is just not going to happen over night. I agree with you in well managed IT shops, locked down etc it's easier but even then we should not underestimate what the tools can really do to handle individual users. You only need one exception to break that model, and at scale that happens a lot, destroying the TCO of VDI. Therefore Layers technology is a way forward to address that. This is why when I speak to many people who are implementing VDI, they are 99% going 1-1 persistent images and pooled in small use cases where there is a lot of similarity. In that case XA is cheaper anyway.

Wow, all that without a curse. Have a good weekend all.


I work at a large firm with global users and have a HUGE Xenapp farm.  The CCU licensing model works great for us as at any given point of the day we have over 10K users logged in with a total population of 60K users globally.  This allows us to buy only 14K licenses (in the case of end of month /quater/year spikes in usage).  We are (or were) looking to take the same approach with XD.  20K licesnse for 45K users globally based on the time zone differences this worked out perfect.  

Going to the Per Named User, which to my understanding is that anyone accessing the XD environment at any time needs a dedicated licesnse, would increase our license spend by more than double.    

If this turns out to be the case when XD is offically release then we need to rethink our deployment and possibly grow our Xen App Farms more and use XD only for VM hosted apps for those that will not or perform horribly on Xenapp


And forgot a big point. This is for XenDesktop. XenApp is still around. If you really love the CCU, just go with XenApp. No one is point a gun in your head to get XenDesktop. Why the heck are you comparing apples and oranges in the first place?

BTW, why are you guys so upset when your choices are not taken away? Just go get XenApp. I know for my company, we are XenDesktop all the way. In fact, we see a ton more benefits and so do our peers.

Good luck with some of you who are so locked in as the XenApp guru. The future is the desktop and if you can't switch, you are going to be out of a job in the future anyway. Just think.


I do find these results skewed and not a correct representation of the customer community.  Almost one third of the participants of this poll are IT consultants and/or resellers.  That in itself does not accurately represent the customer community.  Yes you can say because you speak to customers and you are relaying what customers are saying.  I have been a consultant for about 7 years now in the Microsoft/Citrix/VMware space.  I have many customers who I work with throughout the south central region (I'll let you figure out those states) and the response I got initially from my customers were not favorable because of the abrupt change in the license model.  Once I sat down with these customer on an individual basis and discuss what exactly XenDesktop 4 was going to be able to do for their environment, the perception changed.  Once they realized this is a fundamental change in the way they deploy desktops and applications to the end user then the cost associated with it made more sense and was now able to stomach.  Once customer with 28,000+ users initially said XenDesktop 4 was completely out of the question and are now thinking about buying the product for that user count.  Here is the kicker, they are in one of the two vertical markets that are complaining most about the new license model.  Please don't mistake this post.  I completely agree that Citrix should maintain a CCU model for both XenDesktop and XenApp because the per user model does not completely work for everyone.  If they provided both per user and a CCU model I do not believe VMware or any other competitor would stand a chance in this space.  That being said, keep in mind what exactly XenDesktop 4 can do for your entire user base before you fall in line with the rest of the people who do not view change appropriately.


Citrix could have easily avoided this mess if they had just added the per user license as another licensing option to CCU.

I agree with everyone who says that the shift to per user is right. But I think everyone can agree that Citrix could have done a much better job at selling this change to its large customer base!!

Flexibility and choice is important not only in the technology to deliver VDI but also in structuring costs for VDI solutions.

I hope this whole exercise was a bit of a wake up call for Citrix. (Though I doubt that it really will be)




You talk about those who are here participating in the discussion as those who "do not have a wider view of business." yet you use XenApp on a one to one basis.

Don't get me wrong that's totally valid !! But not when you have a "wider view of business".

You aren't getting double license in the conversion, you've probably had double the whole time.



Wider view of business means not only buying Citrix to solve application issues.

It's bigger than just applications. It's bigger than just one types of users. We are talking about service the entire corporation. CCU is dead. Not Microsoft, Oracle, Apple, etc use this out dated licensing model.

Like I said, go XenApp if you want CCU. That's not going away. Don't complain that Xendesktop doesn't have CCU any longer. It's a huge mistake for Citrix to have stuck around with CCU so long.


@rebecca I think you are missing the point.

It's got nothing to do with understanding the value of the Desktop OS. It's BASIC math. If you are not sold on all out VDI, per-user WILL cost you more. Therefore forcing a new license model on all and saying F it is the sign of a naive organization and stupid management team.

Just look at what MS did with SQL server when they went from processor based to client access. For a while you could buy both. That's what happens when you understand the market you serve. Clearly not always the case with MS, but in this case a good example.

In the case of Citrix just a stupid XD mgmt team who are trying to force licensing and architecture on the world. Just like those idiots try to force a broker when it's not needed in so many cases when nobody can implement pooled for years and nothing they are doing is adding value yet.  They need to understand the world will move to per-user in their own time not Citriix's. Therefore it is wise to give people choice and at the same time encourage the move but NOT force it.


I just spoke to a Citrix sales rep and he just said that Citrix is about to backtrack on per user licensing and an announcement will be made shortly on what it will be instead :)


@ Rebecca It comes down to simple math with licesning.  At any given point in the day we wouild have 20 to 30K users logged in using XenDesktop with a user base of 60K users,.  With CCU licensing I would only need 30K licenses not 60K, therfore with CCU my licneseing cost is half.

And you statement in reference to Xen Desktop being the future, Do the math  300 Concurrent users running office,task and KB type apps on a single server (XenAPP) vs 6 to 7 Physical Servers running 300 Desktops which one is cheaper .......

You will always have more economies of scale with XenApp / terminal services then you would with a Desktop.