Microsoft is loosening the reins on Vista licensing in an effort to let businesses try out some new computing possibilities, including "diskless" PCs.
For the most part, the Windows licensing terms have assumed that the whole PC is going to be in one place; however, increasingly, that's not necessarily the case.
Virtualisation technology means that one physical computer can act as many separate computers, while higher-speed networks mean that different parts of a computer can actually be housed in various locales. For example, it is now possible to have a diskless PC, in which the main hard drive of the computer is actually stored in a data centre, while all the other parts — processor, graphics chip and memory — remain at a worker's desk.
But until Sunday, there was no proper way to license Windows for such a computer. Under new licensing terms for Windows Vista Enterprise, businesses will be able to use the corporate edition of the operating system to handle this as well as other niche cases in which a PC's storage, computational power or both are handled somewhere other than the desktop.
"We're responding to enable a set of early adopters in finance and governments, in particular, to take advantage of architectures that centralise Windows," said Scott Woodgate, director of Windows Business Group. "They either centralise the storage of Windows, the execution of Windows, or both, in the data centre."
In addition to the diskless PC, the other arrangement Microsoft is giving the nod to is one in which a desktop PC or thin client is running multiple virtual machines that are running on a server. The Vista Enterprise licence already provides a licence for additional Windows virtual machines, but only those running on a desktop PC, not for ones taking place on a server. The new approach, called Vista Enterprise Centralized Desktop, requires an additional annual fee.
Although Microsoft is making such set-ups legally permissible, it doubts diskless PCs will become an overnight trend. There are many negatives, most importantly the need for a constant and uninterrupted high-speed network connection. If the network goes down, the whole PC becomes unusable. Also, while Microsoft is making diskless PCs commercially possible, enterprises will still needed additional third-party software to actually make such systems boot up.
Because of the technical limitations and the massive IT resources needed to manage such an operation, Microsoft expects only a small number of institutions to try out such a set-up, most likely top-secret government agencies where security concerns trump the inconveniences.
"If you are in a department that is a three-letter department you may want to keep that hard drive away," Woodgate said, likely making reference to organisations such as the CIA or FBI. "It's relatively a niche."
Even in those cases, Microsoft says it will take time to get the systems up and running. "It will be interesting to see, 18 to 24 months after those early adopters have taken the systems into production, how successful they are," Woodgate said.
Talkback
I have installed a 30+ diskless PC environment and I can tell you that it most certainly does NOT have "technical Limitations" (BERYL 3D desktops are loved by the users), nor does it require "massive IT resources" (just 'lil old me), and it is a major convenience as opposed to an inconvenience (centralised management is a snap)... but then again the set-up of which I speak runs Fedora Core 6 workstations diskless booting from a CentOS 4.4 server.
The single point of failure is the server, true, however since it is NOT Windows if the server goes down another machine can be slipped into place, the old hard disk moved (or a backup image restored) and CentOS will simply restart. It won't think it is being stolen, it won't insist on you entering a long-winded licencing key or having to phone the vendor to reactivate the server. It will simply reconfigure itself and keep on chugging. Next up is to set up a high-reliability cluster as the server, at that point the single point of failure would be the electrical company and how long the UPS can hold out.
Now THATS service.
You know, for a 'niche' concept, virtualization sure is a convenient and commonly used setup.
It's only a niche in the minds of the Microsoft management who seem to think anything they haven't personally trumped as an advantage isn't worth looking into.
Just ask VMWare how successful the virtualization business is.
I look at this in terms of security updates, in fact any kind of updates, and it makes me shudder. The relationship becomes a many-to-one, instead of a citrix farm setup of one-to-many.
What am I talking about? Well, in Citrix, you have a machine that services many clients.
In virtualisation, you potentially have a client that gets served by many machines.
Its this inverse relationship that really gets my hackles up. If you had a citrix farm with 10 servers for 500 clients, who were running a wise terminal, then you only have to patch a small number of devices. This is the benefit of a centralised computing model.
On the other hand, when you introduce virtual computing, every business department wants to have a host for every application, and you typically end up with a huge number of devices that need to be patched.
And patching them when they arent spun up has yet to be achieved in any enterprise sense.
So we come to that oft-heard, and frequently mis-used term, of total cost of ownership.
The effort that large companies go to to ensure a robust and uptodate environment is nothing short of incredible. And I cant help but think that better design up front would have prevented much of this. And now we have this virtualisation coming, and it seems to me to be a disaster in the making.
Not a fan.
Citrix is also a violation of the KISS principle.
It's sold as a KISS solution but in reality it isn't.
For one, you're never going to get a dumb terminal only environment. Employees will still bring in laptops, PDA's, require Site-to-Site access, demand desktops for specific tasks, etc, etc. That all will need patching as well. And even Wise Terminals need updates from time to time.
Another thing is that Server Based Computing (SBC, like Citrix) is a whole different ballgame. You just don't go updating such an operational environment without going through your OTAP procedures. And that takes time to do it right.
Indeed, a better design upfront is required to prevent a lot right from the start. However, such a design would be based on the KISS principles. So no 'all your eggs in one basket' or 'consolidating and oversimplifying the heck out of contingency, choice, flexibility and availability' in such a design. Just for the sake of cutting down costs, on paper.
While I agree with many of your comments, the fact is that Citrix, and lets step back a moment and say a centralised computing model, are easier to manage than a large number of machines that are switched off, or unavailable for many reasons, such as user tampering, remote locations with poor network links, and so on.
If I had the choice to maintain a small number of big machines, versus a large number of small machines, to me its an easy answer.
But each to their own.