You may want to mark your calendar – one week from today, Microsoft is holding a “Desktop Virtualization Hour” Webcast, beginning at 9:00 am Pacific time.
Why should you care about this? Well, techtarget.com and others are reporting that there are some licensing changes coming that will make it easier – and hopefully less expensive – to license desktop virtualization technologies.
I really hope they’re right, because the current license model is complex, expensive, and, in my experience, not very well understood by the user community. The vendors of VDI technologies are often not very helpful in this regard, because their focus is to sell you on why their VDI approach is the best, not to needlessly (in their view) complicate the sale by saying, “By the way, here are all the hoops you have to jump through to legally license this deployment from the Microsoft perspective.” Not to mention the fact that their salespeople may not understand the Microsoft licensing side very well anyway.
Today, there is one and only one way to legally license access to virtual Microsoft desktop operating systems, and that’s with the VECD license. (When it was first introduced, VECD stood for “Vista Enterprise Centralized Desktop” – today it stands for “Virtual Enterprise Centralized Desktop.”) The VECD license is only available as an annual subscription license…and the annual cost varies, depending on what the client device is that’s being used to access the virtual desktop OS. If the client device is a Windows Desktop that’s covered by Software Assurance, then adding a VECD license for that client is only $23/year. If it’s anything else – a Windows desktop that’s not covered by Software Assurance, a thin-client terminal, a Linux desktop, a Mac, etc. – the cost jumps to around $120/year. That adds up pretty quickly, and it goes on forever. That’s a tough license model to adjust to if you’re a Small- or Medium-sized business that’s used to just buying new PCs that come with OEM copies of Windows.
These days, you can buy a pretty darned good desktop PC with a professional version of the Windows OS, and probably an OEM copy of the Office Suite, for $700 – $900. If you amortize that over three years, well, you can do the math. Throwing another $120/year on top of that for the VECD license is not insignificant…and we haven’t started talking about the cost of the rest of the VDI infrastructure.
Don’t get me wrong – I’m a fan of VDI in general, and Citrix XenDesktop in particular, which you already know if you’ve followed this blog for any length of time. I truly believe that there are overall cost savings to be had. But most of the savings are in soft costs: reduced effort to manage the desktop image; easier to upgrade and patch; harder for users to break things; easier to insure that critical data is being stored in the data center and backed up; flexible application deployment options; consistent access to the same desktop from just about anywhere; extended life for the client device; faster rollout of Windows 7; etc. Those cost savings are real. Unfortunately, they are also, by definition, hard to quantify, despite the best efforts of the Gartners of the world. And a lot of businesses are still in an operating mode where reducing hard costs today is more important than reducing soft costs in the future.
Reportedly, one of the changes Microsoft is announcing will be a move from per-device licensing to per-user licensing. Depending on what the numbers look like, that could be a step in the right direction.
The other change I’d like to see – and this applied to Windows Server licenses as well as to desktop OS licenses – is to have the OS license associated with a workload, not with a physical device. If a Windows Server license was associated with a workload, e.g., my Exchange Server, instead of me having to “assign” it to a piece of hardware, then I could use live motion to move it from one virtual host to another, or use HA functionality to restart it on another virtual host in the event of a host failure, without having to worrying about whether I’m violating my Microsoft license agreement.
On the desktop OS side, I tend to agree with Tony Wilburn of Betis Group, Inc., who is quoted in the techtarget.com article as saying, “If I buy a Windows 7 license…let me use that instance of Windows 7 whether I have it installed locally, attach to it remotely with a PC or thin client running Windows or Linux, [or] have it running on vSphere, XenServer, or Hyper-V.”
Microsoft’s answer to the Server OS issue is: buy Windows Datacenter Edition licenses for your virtual hosts. And that does indeed solve the license compliance issue. It can also nearly double the cost of licensing for a smaller business that is looking at server virtualization for the first time.
So far, Microsoft’s answer to the desktop OS issue is: buy the VECD license.
But Microsoft is not totally insensitive to customer pushback. Some of us remember when NT 4 Terminal Server Edition was first introduced. Microsoft’s initial licensing stance was to say, in effect, “Terminal Services provides another way of accessing an NT 4 desktop. Therefore, if the client device you’re using is not an NT 4 Workstation, you must buy an NT 4 Workstation license for it, no matter what it is.” Customers who were running Windows 9x, or thin-client devices, or Macs, were not happy about being told that they had to buy a bunch of NT 4 Workstation licenses that would simply sit on a shelf or in a file cabinet and never be installed. The customer outcry was so loud that, less than a year later, Microsoft converted to the Terminal Services CAL license model – which at least had the virtue of being consistent with the way they licensed other server products.
Windows 7 adoption is important to Microsoft. And Windows 7 adoption is driving a lot of the current interest in VDI. Therefore, it is in Microsoft’s best interests to make it as easy as possible for customers to deploy VDI as a means of enabling and accelerating Windows 7 adoption. The signs are hopeful, we’ll just have to wait and see what comes out of next week’s Webcast.