Think Citrix XenDesktop is cheaper than XenApp? Don't forget about Microsoft OS licensing costs.

This blog entry was written by Rob Hammersmith, a Citrix sales engineer in the US central region. He wrote it in response to some blog items I posted over the past few weeks.

This blog entry was written by Rob Hammersmith, a Citrix sales engineer in the US central region. He wrote it in response to some blog items I posted over the past few weeks. Rob allowed me to post it here as long as I made it clear that these are his personal thoughts and not the official line from Citrix.

Recently Brian Madden posted a couple of articles on his website regarding the pricing of XenDesktop in regards to the pricing of XenApp. In the most recent article he states, "Too bad the CXOs of the world are going to read that XenDesktop is the 'cheap' way to XenApp and want to deploy it for all their users immediately."

Well the CXOs of the world are going to read that XenDesktop is the 'cheap' way to XenApp BECAUSE Brian keeps writing it. But just because Brian keeps writing it doesn't mean it's true.

For a fraction of customers it is true, but for most customers it's not. In fact, some customers could end up paying two to three times much just in licensing costs alone just to do XenDesktop in lieu of XenApp, despite a concurrent XenDesktop license being cheaper than a concurrent XenApp license.

Why? Because you can't build a XenDesktop solution (nor a XenApp solution for that matter), without paying Microsoft and hardware vendors in addition to Citrix. While Citrix charges less to do XenDesktop than XenApp for the customer to actually do XenDesktop rather than XenApp, it could costs significantly more overall.

I'm not going to go into the increased hardware costs for XenDesktop, as that's already been hashed out in previous responses to Brian's articles. What hasn't been discussed much is how much Microsoft makes in a XenDesktop solution (or any VDI or Blade PC solution for that matter).

We're all familiar with how Microsoft makes money in a Terminal Server solution. For every Terminal Server you stand up, you pay Microsoft for a Windows Server license. For every client (or user) that connects, you pay Microsoft for a TS CAL. Simple enough.

But with XenDesktop, people assume it works the same as a licensing a TS solution. It's doesn't. Sure, the VECD looks like an equivalent to a TS CAL. And at $23 MSRP, that's a lot cheaper than a $119 TS CAL. One could look at that and say, "if I have 300 clients, Microsoft gets $6900 ($23 x 300) if I choose XenDesktop and $35,700 ($119 x 300) if I choose XenApp. XenDesktop is a no brainer, right?"

Not so fast. You have to remember that both a TS CAL and a VECD are just a license to remotely connect to a licensed operating system—they don't actually license the operating system. For TS, the license for the operating system is done by buying Windows 2003 and installing it on the server. If I need a second TS because I want to load balance those 300 users across two servers, I have to buy a second Windows 2003 server license, (and so on and so on). So for 300 users I'm looking at 300 TS CALs and a few Windows 2003 server licenses.

But in XenDesktop—and any VDI, blade PC, or remote single-user Windows solution out there—the operating system still needs to be licensed. But it's not like TS where one operating system license is installed on a server that's shared via a pool of TS CALs that client devices check out as they connect. Microsoft ties VECD Software Assurance (SA), which is tied to a Vista OS, which is tied to a client device. That means for those 300 users to be licensed by Microsoft for XenDesktop, Vista Enterprise has to be purchased 300 times. And then with 300 upgrades to SA, I'm eligible to buy those 300 VECDs (assuming you're in the right Microsoft open licensing agreement to get Vista Enterprise, which is the only edition of Vista where VECD is available).

So with XenApp, I buy 300 TS CALs and a few Windows 2003 server licenses. But with XenDesktop I buy 300 Vista licenses, 300 upgrades to SA, and 300 VECD licenses.

Now a good argument can be said that some customers are already there, as all their clients may already be licensed for Vista. But the differences don't stop with the initial purchase. They keep going.

In a TS environment, both the server OS license and the TS CALs are one time purchases. I can use the solution as long as I want. That's not true with VECD. If the SA and VECD are not renewed annually, then that client is no longer licensed for remote connections to Vista (or XP). So even customers who already have Vista licenses for all their clients will eventually eat up all the savings XenDesktop provided over XenApp in year one through required Microsoft annual renewals (Citrix has SA renewals too, but they're optional, and the cost of Citrix SA is the same percentage with either XenDesktop or XenApp).

Now Microsoft does have another option, which is the VECD for thin clients. With that all the client needs is the VECD (no Vista purchase or SA upgrade necessary). But instead of $23, it's $110. So back to our scenario, if I have 300 clients, Microsoft gets $33,000 ($110 x 300) for XenDesktop and $35,700 ($119 X 300) for XenApp. XenDesktop is a no brainer again, right?

Wrong. This VECD for thin clients is annual too. Don't renew it, then don't ever connect with that thin client to Vista (or XP) again. After a few annual renewals of these VECDs for thin clients, the savings XenDesktop provided over XenApp in year one is again eaten up through required Microsoft renewals.

But wait, that's not all. What about the TS environments where the customer has assigned their TS CALs in "per user" mode, like if they have 100 users that have 300 devices they use to connect with? In that scenario, with TS I just need 100 TS CALs. But with XenDesktop I need 300 VECDs. There is no user mode licensing for VECD. Why? Because again, all the licenses (the Vista OS, the SA, the VECD) are to a specific devices. There's no way for Microsoft to tie it to a user (other than having a named user for the device that all the Microsoft licenses are tied to, but that's still a per device mode). Using XenDesktop in lieu of XenApp in this scenario is a money pit.

And of course there are those customers that look at this and say, "Well, I'll just avoid that whole mess and stick with doing VDI with Windows XP." Sorry, Microsoft doesn't allow that. Well, they do, but to be licensed for remote connections to Windows XP you still need the VECD license. And the only way to get that for the client device is to buy Vista Enterprise for the client device and upgrade it to SA. You can then use your "downgrade rights" to use that VECD as a license to remotely connect to Windows XP. So sticking with Windows XP and XenDesktop is not a way to avoid all of these Vista, SA, and VECD license costs.

Tthe bottom line is there are some customers who could likely save money for a few years choosing XenDesktop over XenApp. These are customers who already have all of their clients licensed for Vista with all of them are upgraded to SA, and doing XenDesktop (or any VDI, or Blade PC solution) is just an effort of adding an annual VECD cost to their environment. Now whether the money they save covers the additional hardware they need to do VDI over TS is another debate all together.

For those that need more than just a simple VECD purchase (because their Vista is not on SA, or they aren't licensed for Vista yet), odds are they will pay much more for XenDesktop over time than they will for XenApp, even if they happen to save money in the first year or two. Until Microsoft either releases the annual requirement on remote connections to Vista (and XP), or forces TS CALs into an annual subscription as well, this won't change.

Ultimately all of the comparisons above assume that XenDesktop or XenApp would fit the customer's needs. And if this is true, the customer would be better off using XenApp, as with ongoing licensing costs, and additional hardware needed, XenDesktop is more expensive. But if XenApp doesn't fit the customer's needs and XenDesktop does, then why even debate which one is cheaper? Just buy XenDesktop.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

This is all great and all, but it sidesteps one of the main reasons many look at VDI - a great many of the users that are using applications from a Terminal Server STILL have a desktop.  Companies are already paying for the OS license and the TS Cal and the TS OS.  They just need to purchase the VECD and move it into the datacenter to implement VDI.  Most of those larger companies also already have/are paying for SA.... so I just can't see how the arguement above has any merit whatsoever is the vast majority of environments.  Maybe the SMB folks?

Show me a full-client client environment - full - 80% or higher of the users with no fat client; and then the discussion above has merit.  That just doesn't seem to be the case out in the field.

There is a REASON the whole IT world isn't only server based computing and all thin clients everywhere.  And now, companies perceive that they are obtaining many of the advantages of SBC without all the perceived limitiations by going to VDI.  Right or wrong, it is.

It seems to be that the smart thing to do might be to go for Provision Networks's offering that offers the benefits of SBC with the option of virtual desktops for those users whose requirements exceed the functionality of SBC (in reality, I would suggest in many organisations this latter group is smaller than people imagine). It costs less to purchase than the associated licensing costs of Citrix.

The blog isn't really applicable to the context of companies that are looking to add a VDI/Blade PC solution in addition to an existing TS solution. In fact, most enterprise sized environments are going to have both. The considerations of doing a VDI/Blade PC solution AND a TS solution are different than what's posted above.

The blog is within the context of a customer who is considering doing a VDI/Blade PC solution OR a TS solution, and determining which solution is cheaper is a major decision making factor.

And the "OR" debate isn't a new argument that I've brought up, as the hardware considerations of each solution have already been hashed out here on Brian's site and elsewhere ad infinitum. I'm just extending the argument to other considerations. The post above is just pointing out that licesning a VDI/Blade PC solution with Microsoft is far different than licesning a TS solution with Microsoft, and the customer needs to consider where they stand with Micrsoft licensing today and how the costs play out over time in order to really determine whether or not XenDesktop is really cheaper than XenApp. 


You will need all Microsoft licensees anyway. You can use whatever VDI broker, you still need VECD, OS costs etc. Provsioning Network allows you to lower non-Microsoft costs only. However I totally agree with your opinion: VDI does not bring any advantages for most of the users over Citrix SBC solution. User needs the application, not desktop to work. We have Citrix SBC now here in our company, we do not need streaming, Access Gateway,VDI ...  . All this stuff may be usefull, but usually you are O.k with presentation server.


If I only have a hammer, everything look like a nail...

A desktop with applications inside or an applications on a desktop, even if it look like the same (dekstop+application) are not identical... in this case a+b is not equal to b+a.

If you add the fact that 2 identical companies (same size, same business and same IT infra) will not have the same need regarding IT evolution and will not make the same choice based on different histories, knowledge, skills, marging structure, strategy and budget (cash or not)...

AND or OR is a theoritical discussion for which everybody will have their own pro and con.

Instead of trying to justificate on eor the other solution, why not make a complete and clear matrix (like is but more colomn including "best technical target", price structure and implication, ...

anyway, always good to read toher point of view ;-) 


Must be desperate to sell licenses for Vista.  If you can't sell it on merit, force people to buy it.
This is a point I have been trying to get across on the Thin list.  Unless you have the desire to spend a whole lot of money the VDI model is not always the solution. You could expect to pay around a million dollars for 1000 users after you purchsed servers and licensing. This doesn't even include pricing for the client devices and OS.
One of the big sales lines from Citrix is that XenDesktop can be used from any device anywhere. However, if the VECD license is "per device", how do I license use from the airport lounge, the hotel and from my users' home PCs?

Kudos to Rob for sticking his neck out and telling it like it is! Vendors sometimes forget how important honesty and integrity are when trying to gain a prospect's trust - it's clear that Rob hasn't and I for one appreciate it.

I have been reading about VDI for months now and I'm still trying to come to an informed opinion. The amount of sales spin and misinformation is frightening. I can't imagine how someone new to VDI and/or SBC makes any sense of it given the fact that we in the SBC community are still digesting and debating the facts about VDI vs. SBC.

The more I read about VDI, the more I realize that it's appropriate only for those circumstances where SBC isn't suitable. SBC adheres better to the KISS principle than VDI, and I don't buy the argument that SBC is more complex than VDI as administrator's of VDI solutions need to contend with virtualization and storage challenges in addition to the management of VDI instances themselves.

VDI might just be the worst thing that could have happened to SBC in a long time. What incentive is left for Citrix, Provision, et al to improve this technology with everyone falling over themselves to sell VDI solutions.


Hmm, guess I'm starting to form an opinion about VDI afterall :-)

I've been advised by Microsoft that VECD extends to home user rights, so your users can access a corporate virtual desktop from their home PC. If you want to access it from an airport lounge or hotel, you're buggered - another plus point for SBC solutions.
Good point about the additional virtualisation and storage challenges of VDI. I would suggest though that the SBC market is the victim of Citrix's monopoly of the technology for 10 years. Put simply, Citrix had no reason to blow the market away with their technology because their was no real market to blow away. None of the major software vendors acquired Ericom or Tarantella. All Citrix had to do was keep ahead of terminal server. Once they had ICA, the race was largely won. As a result, their core product is still a bit flaky after 10 years. VDI is benefiting from three large software vendors really going for it - Citrix, Vmware and Quest, with Microsoft watching closely.
Thanks, Rob, for this very informative article and for putting things into perspective!

Removing the VECD is what will make VDI a reality. If I pay for a Vista/XP License-it should pass all the way through to Active Directory no matter what device or location I choose. That is a true end-to-end cloud. otherwise it's just smoke and mirrors and M$ wins again!

My opinion, not companies. R Wheeler (somewhat new Citrix Admin, 20 years in IT)


Alan, any improvement we make to RDP, Seamless Windows, Application Publishing, the Provision Management Console... is done with both Terminal Services and VDI in mind.  The only features we make that are particular to VDI are the integration with and automation of the Virtual Infrastructure.

As for the inferrence that VDI Administrators need to manage storage and virtualization, this just depends.  At many customers I meet with, there are Citrix Administrators, VMware Administrators, Server Admins (Microsoft or Open Systems), an Application Packaging Team, Storage Architects, and Desktop Support Engineer, or some lesser combination thereof. 

What is happening is that the Virtualization Admins (VMware, Virtual Iron, Hyper-V, Virtuozo, XenServer....) are adding more hosts, the storage people are adding more capacity and the Desktop Support Engineers are handed a new console and easier way to manage their environment.  Nowhere have I seen where the Citrix admins have taken over management of desktops (from the Desktop Support Engineers), as this doesn't make financial sense, as a Citrix admin makes $75K-$125K and a Desktop Support Engineer make quite a bit less (no offense intended). 

The reality is that most IT Departments at large corprorations have dozens of engineers of some skill set, but one a select few that have any clue whatsoever about Citrix and Terminal Services.  Moving a larger majority of application delivery and management to the desktop OS accomplishes a few things:

1.  More IT staff are able to effectively support the end users, if applications are being delivered by some technology they don't understand (TS/Citrix)

2.  No absolute reliance on custom application packaging for a Multi-User OS

3.  No worry about printer drivers being multi-user aware

4.  No worry that applications won't work on a multi-user OS, as VDI is run either on XP Pro, Vista or Server 2003 (Virtuozzo).

5.  No worry that doing something for one user is somehow going to adversely affect everyone else.

Is VDI Simple to implement? No.

Is migtration from TS/Citrix to VDI simple? No.

Is managing a Virtual Desktop simpler than a Physical Desktop?  Absolutely.

Are there going to be hybrid solutions? You betcha.

Are people that are not deeply invested in TS/Citrix going to deploy a hybrid solution?  Not that I've seen.  They only people I have seen that are considering a hybrid solution, are those that have capable TS/Citrix Admins with a large investement in that technolgoy.  Those that do not have particular expertise with Citrix/TS, or don't like managing such an environment are eager to implement or migrate to VDI.

As with anything, there are not absolutes, as every business and IT Department is a little different.  What is a great fit for some, may not be a great fit for others.

The cost of the hardware and software are historically reported as less than 30% of the TCO for a desktop, whereas the maintenance and support of that desktop is nearly 50% of the TCO.  If that 30% changes slightly, but the 50% reduces dramatically, you're saving money.


Nice article, but what if we replace the VDI desktop with something like say Ubuntu Desktop? That would need no license! So then all those M$ licenses can be done away with :-)

Ofcourse this depends on the apps one uses within the VDI desktop, but hey it's just an idea 

I know that the Ubuntu Desktop uses VNC now to connect remotely to a desktop, which is far from ideal. But who knows, that might change in the future into a real remote desktop protocol! 

Excellent.  This is why this site is the best Citrix site out there.  Keep it coming. 

If you want to hear why one company is putting together a VDI solution..

We are ramping up to support several hundred external developers and application support personel via static VDI. They all use a common set of tools and utilities, but each has their own unique development utilites they employ. We have TS experience and 150+ TS boxes. A developer needs a desktop to be productive, and when paying by the hour you want them to be productive. So, VDI and VECD/Thin it is. Once this demand drives the initial infrastructure creation (the chicken has arrived), then other niche uses within the company (the egg) become easier to fill. As a central IT group, we can offer centrally backed up desktops with full DR capability. We can support SOX eDiscovery requirements. We can offer short term / interim solutions for peak needs (consultants onsite to work for a few weeks who need a machine) with instant delivery and no re-purposing issues when the need is over. Once we go with dynamic pools, we can offer "free seating" which we do not have set up now. I don't expect anything to take over. I look forward to new options for our clients.