There’s been plenty of fanfare surrounding the release of Citrix XenDesktop 4.0 this week. I plan to blog about the XenDesktop 4.0 technical features after I’ve spent some quality time with them in my lab over the next couple of weeks. However, I’d like to contribute to the XenDesktop 4 conversation by talking about its impact on client software licensing.
It’s no question that user expectations for how they should access corporate resources are changing. Sometimes I need to quickly grab or view a doc from my iPhone. Other times I need to view resources on my laptop while not connected to any network. And of course, most times I’m in my office and can access my applications while fully plugged in. As I see it, I’m just scratching the surface. Thin or zero clients, netbooks, and remote Internet kiosks (such as at a hotel or conference) are increasingly becoming part of the the application access picture.
So what does this all mean? Users just want to get at their data and applications on their terms. IT should be able to provide that level of service, and from a technology perspective we’re getting there. Sure I’d like to see capabilities for deploying endpoint security to devices such as iPhones before considering them enterprise-ready alternatives for application access (and a Bluetooth keyboard wouldn’t hurt either). However, for all the gains we’ve had in technology, client application vendor licensing and support still remains one of the primary barriers to wide scale desktop virtualization deployments.
Microsoft, for example, still licenses desktop OSs and applications by device, or installed instance. Sure the Vista Enterprise Centralized Desktop (VECD) licensing model includes a “home access” provision that allows users to access their virtual desktops from their home computer. However, this licensing model still expects the organization to count devices. I’ve talked to Microsoft about my concerns over per-device licensing and for the most part, we are in full agreement that Microsoft desktop OS and application licensing will need to fundamentally change. However, for an organization as large as Microsoft, this is going to take some time. Any licensing change has a huge impact on existing OEM and sales channels, and is why licensing changes are often incremental in nature.Still, Microsoft has an opportunity to lead the way and show other vendors how to license software for today’s increasingly mobile user. I hope they embrace that opportunity.
The user experience is moving toward an era where user data and applications live in the cloud. In other words, the cloud is their desktop. Sure this could be a server-hosted Windows XP or Windows 7 instance or something different (e.g., client-hosted desktop, virtual applications, or a mix of applications and services delivered by internal IT and external PaaS or SaaS providers). The bottom line - the way we deliver applications and services to users is fundamentally changing. Now is the time for the vendors to define policy that meets the needs of how their users will access applications. I believe the best model for the emerging virtual desktop and application delivery methods to be a per-user model. The point of virtualization and cloud is to abstract (or decouple) the physical dependencies of IT services and applications. We’re very close to being able to seamlessly achieve this with technology. With many organizations planning major desktop virtualization rollouts in 2010, it’s time for application vendors to rework their licensing models. I am advising clients to draw a hard line on licensing requirements when they put prospective vendors through the RFP process, and I advise you to do the same (i.e., require per-user licensing).
Reworking licensing to be user-centric needs to be a top priority among client application vendors heading into 2010. Vendors that insist on binding licenses to physical devices in an increasingly virtualized world are not part of the solution. They’re part of the problem.
#1 by Joe Shonk - October 7th, 2009 at 12:34
Are you saying that Schools and Hospitals should pay per user instead of device? This would be prohibitively expensive.
For a school, instead of paying 50 Devices License for 50 PCs in a lab, I have to make sure that I have a OS and App License for every student that will use that lab? Ouch.
If I’m at home then I need 6 licenses of Office for my home PC so the kids can do their homework?
Joe
#2 by Chris - October 7th, 2009 at 12:55
Thanks for the feedback, Joe. Let’s keep in mind that licensing is relative, and vendors can adjust a per user model price so that their revenue streams are protected, and the customer is not paying more for the same service. Similar adjustments occurred when vendors dropped socket-based pricing in favor of virtual CPU or instance-based models for server virtualization environments. Once we start mixing in client hypervisors and reach a point where the endpoint is nothing more than an access device, the model has to change. Granted, we’re not there yet - far from it. Seeing through layers of virtualization to associate an app with a physical device is becoming increasingly complex, and is something I don’t see as sustainable. To me, the per-user model can work, regardless of platform. There are issues with the home user market that need to be resolved too, but I think Microsoft’s forthcoming free version of Office in the cloud speaks to the trend.
#3 by Joe Shonk - October 7th, 2009 at 14:21
Ok, I buying to the fact that vendors can tweak pricing but how do they tweak the value provided to the name-user? A user that uses the application all day long will receive the full value of that license. Users that access the same application on a part-time do not receive the full-value of the license? What if the user (say a Nurse) works for multiple employers. If the license is per-named-user the vendor gets to charge multiple times for a single user.
Joe
#4 by Daniel Feller - October 7th, 2009 at 15:30
Chris
What you will find is that no matter what the licensing model is, about 50% will be unhappy. For example, Joe doesn’t like the per-user licensing because in his example there are more users than there are devices. Pretty common, especially in schools. But there are also many other examples where the opposite is true. You have more devices than users. So in this case a per user licensing is the better option.
What’s the solution? A mixed bag? Pick which version you want? Probably, but will anyone go down that route? Doubtful.
Twitter: @djfeller
#5 by Chris - October 7th, 2009 at 15:57
Daniel - I agree. In most cases choice is the major issue (to Joe’s earlier points). Server app vendors offering choice of models have fared well, and I think the same should be expected for the client app vendors. As we start getting into scenarios such as employee-owned PCs (granted - not a huge issue with public schools today) complexities associated with device tracking and binding licenses to devices will only grow worse. Choice is key. And as Daniel states, satisfying the other 50% can be supported with per-user or concurrent user licensing models.
#6 by John Lockie - October 7th, 2009 at 17:15
I hate per device licensing. It’s a lose lose when you really step back and look at it. I run IT for a traditional corporation, and I cannot wait to see 100% of our licensing go to a user basis. It would help me manage TCO and cost per employee for IT services a LOT better.
The only places per device works is where multiple people access one machine continually. In any case, developers and vendors are not so stupid to charge the same per “user” as they would per “machine”.
Also, why hasn’t anyone mentioned that licenses for users don’t have to be “hard coded” to a user. You can pool user licenses much like you do, for example, DHCP leased addresses. So if you are a library with 10 machines, buy 10 “user licenses” and you’re done. People are not realizing you can’t really have more then 10 USERS on 10 DEVICES anyways. This way, we are getting maximum use out of our licenses.
On the flip side, it’s more common for users to access services from MULTIPLE MACHINES. I have users licensed on laptops, home PC’s, office workstations, and blackberry phones! What a waste of money. Let me license that 1 user for their set of applications and VM and we are off to the races!
Per user is clearly a necessity. Nice article….
#7 by Shawn E - October 7th, 2009 at 19:52
We definitely need to move away from the per device model, but not necessarily to a per user model. Maybe in application licensing, but in OS licensing what needs to be done is remove the requirement to assign a Windows OS license to a physical PC and assign it instead to an ‘instance’ of Windows.
Virtual or physical - doesn’t matter.
We can then move to a Virtual Desktop Infrastructure where we decide how many virtual instances of Windows we want to run to support the userbase. This allows a user to use a corporate device in the office to access their virtual desktop and then access that same virtual desktop from home or on the road when they leave. These people aren’t accessing a different corporate PC, they’re accessing the same one, so why pay again?
But this provision of licensing per ‘instance’ also addresses the academic or hospital use case where you have more users than PCs, you create as many Virtual instances as you need to support the simultaneous users. In these environments only a certain amount of users are working at once (shifts, classes, etc) and even if some want to work from home when they leave, not all of them do at the same time, so you might only need to create 20% more desktop instances. Whatever the number of additional virtual instances you need to create, it will be less than the user count.
Bottom line is this should be CUSTOMER need driven, not vendor revenue stream driven.
I never understood why OS licenses have to be tied to a physical device - a Virtual Machine running on a Hypervisor is 100% the equivalent of a PC. It can only provide the same functionality as running on physical.
One license = one instance would defintitely make for a good compromise between the two ‘competing’ licensing models and more than anything it just makes common sense.
#8 by Mike - October 9th, 2009 at 07:48
First off, stop referring or inferring to Citrix’s CCU model as per device as its not. Also everyone keeps referring to utilizing of virtualization as mostly an internal service, virtualization has its roots as remote access solution and then healthcare bought into it as it was a way to deal with escalating IT costs. Virtualization was not so much as a destination as it was really a means of saving money. Which the new license model for the above two mentioned deployment scenario’s is about to come to an abrupt end.
Some would say stick with XenApp, give it a rest! XD is a much easier environment the provides a balance between having virtualization, but also remedying long standing application compatibility issues, resource conflicts, multimedia needs, etc.
So the world of the Citrix customers is NOT as simple as many of you would smugly believe it to be. In my case, my Citrix environment cost are about to balloon by seven figures and that is all that matters to this Citrix customer.
#9 by Chris - October 9th, 2009 at 08:35
Thanks for the feedback, Mike. Nowhere in the post did I refer to Citrix CCU licensing as per-device, so I think there may be a misunderstanding. The per-device models I was referring to were with regard to application vendor licensing (e.g., MS Office). There are cases where per-device works (as Joe noted in the comments) and other cases where it falls apart. My post was more about application vendors offering an alternative to licenses that are bound to devices, which is impractical in highly virtualized and cloud-based infrastructures. Virtual desktops with multi-device access are far easier managed with both per-user or concurrent-user models. The distributed user may not always be connected to his applications (e.g., offline VDI), making it impractical for CCU licensing in all scenarios. A Citrix licensing shift will require a pricing adjustment to accommodate for the differences. I’ve worked with vendors in the past where licensing changes resulted in no new net spend among their clients. So the price was adjusted so that transitioning to the new model became a wash. I expect Citrix to take similar strides. Also, I found it encouraging that Citrix offered a survey to solicit feedback - https://www.surveymonkey.com/s.aspx?sm=8KK2ia8zSufDBxMHlyRS_2bA_3d_3d and encourage you to send your concerns directly to Citrix via the survey.
Again, to my earlier point, the per-device model employed by many desktop OS and application vendors is simply impractical in the desktop virtualization space, and new models are needed.
#10 by Simon Bramfitt - October 10th, 2009 at 11:49
Nice post Chris, I wish I’d seen before I wrote almost exactly the same thing last night (thank you :) ). If you look that the licensing question from the vendor end (my heavily revised blog post) you still come to the same conclusion - the only way forwards to move to a per user model, and the best time to do it is now.
Mike - As everyone is only to keen to point out, a change in licensing model is going hurt some, but it is much too soon to determine just how painful the transition might be. Right now it looks like education and possibly retail will be affected more than most other sectors, but there are sufficient remedies available in terms of larger discounts or higher conversion ratios that solution can be found if the vendors so wish. Citrix have already given a clear indication that they are looking to find a solution to this problem. And really if you don’t like the XD4 licensing terms and can’t agree a solution with Citrix - then look to the alternatives. Citrix have sufficient competition that someone will provide you with a solution that matches your needs, you will still need to consider the long term because they are more than likely also move to a user based model in the future.
About the only refuge from user based licensing that I can see surviving will be DaaS based, here at least there might be sufficient flexibility in pricing and format to meet your needs.