If you haven't figured out how to manage desktops in the past 20 years, what'll VDI do?

The #1 reason that VDI projects fail is because companies try to do too much at once-and desktop management is a big part of that "too much."

I've written quite a bit in the past that VDI is about not about making your desktops easier to manage—it's really nothing more than a form factor change that delivers some new capabilities. In our book The VDI Delusion, we talk about how the #1 reason that VDI projects fail is because companies try to do too much at once—and desktop management is a big part of that "too much."

For example, a lot of people buy off on the idea of VDI because it's "easier to manage." They hear about the disk image sharing or cloning or whatever, and they think, "Heck yeah! Managing one disk image is easier than managing hundreds!" And while I won't argue with that, I will point out that you can only actually manage one disk image if you can implement app virtualization, user state virtualization, and a whole host of other desktop environment delivery components. (Of course this is also assuming that all your apps are compatible with app virt, that no users want user-installed or departmental-installed unsupported apps, etc.)

I don't write this to scare off anyone who's thinking about VDI, rather, I just want to point out that if your current desktop environment is one where every user has his or her own unique disk image, and your plan for VDI is that you're going to move to a structured shared image model… you're in for big pain! In fact moving from the personal to shared disk is much harder than the actual VDI technologies.

With shared-image VDI (or any shared-image desktop virtualization), your users are starting off with a gold master image that you customize on demand with application & user virtualization. If that sounds appealing to you, then I have to ask: why aren't you doing this today? If your environment is ready for central image management, why haven't you been using Ghost and SMS and automated app installation for the past twenty years?

If your current desktop environment is physical PCs and laptops that each have their own images with locally-installed applications and you want to move to VDI with shared images—you need to stop right now and create two projects that are unrelated to each other.

One option is to forget VDI and focus on your Windows client management. Think about things like SCCM, Altiris, app virtualization, and user virtualization. Get those all built-out for your physical environment. Only once that's done and successful should you think about VDI.

Or, if you really need the benefits of VDI today, then forget the shared image thing for now. Just recreate your current 1-to-1 persistent image desktop model in your VDI environment. Once that's all up and running you can start to sprinkle in application or user virtualization.

Bottom line: I can't tell you how many people I hear who want to implement shared-image VDI. Then I ask them if they use something like Ghost to re-image their users' machines on a daily basis in their current environment. Usually they say, "no," which leads to me asking why. They say something about how unique their users are or how they have complex apps, and that leads to, "Well, if you haven't figured out how to do this with your physical desktops today, how do you magically think it's going to work in your VDI environment tomorrow?"

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

One reason many have never gotten their hands truly around their desktop management problems is that they have never had the buy-in from their upper management to do so! This is actually something I'm seeing over and over again.

Try to take away admin rights on desktops and implement SCCM in a mature organization, and it is seen as making the users life more difficult and IT's job easier, which gets you nowhere. No sponsorship, no buy-in, and tons of push-back.

On the other hand, if the CIO has a thing for VDI, you can use it as a tool to accomplish some of the management strategies that you have been trying to do over the last 20 years, and use VDI to drive acceptance, sponsorship, and participation.

It may not always be the right tool, but I love to have a long lever to use to crack things open, and in this case VDI might allow me to crack open existing practices enough to really clean out some of the dry-rot that has happened in desktop management practices, and provide some solid replacement technology where needed.

And just think, if you work hard enough at appreciating VDI for what it is (and ONLY that) you can come up with a series called "50 ways to love your lever - VDI as a tool you can use"


Also, good point on App and Desktop Virtualization being 2 different projects, but I challenge the "unrelated" portion. We have had successful App-V sequencing projects running in parallel with Desktop Virtualization with the application sequences begin used as gates to enable on-boarding of more user populations once their apps were available. We actually were able to find more resources and priority for both projects by focusing on the interdependency instead of avoiding it.


Geogres Clemenceau, french general during WWI, homeland minister and "président du conseil" (former president) said once :

"War is a too serious matter to be entrusted to the military."

Isn't it the same for desktops ? too serious to be given to IT services ?


Right on Brian! It's interesting how many people just want remote access to their apps and data...you don't need VDI for that! I've talked a bunch about Citrix RemotePC in the last few weeks leading up to its release and there are a lot of people interested in this. It's easy, cheap, and provides remote access from anything anywhere to a corporate PC people are already using.

IT keeps talking about desktop management and highly managed desktops, users just want their computing to work...I'm not convinced the prior is required for the latter.



"On the other hand, if the CIO has a thing for VDI, you can use it as a tool to accomplish some of the management strategies that you have been trying to do over the last 20 years, and use VDI to drive acceptance, sponsorship, and participation."

Exactly my experience. I work in healthcare and doctors like iPads. Doesn't matter if its the best or even a GOOD way for them to utilize their applications. They are willing to put up with the pain because they perceive iPads as "cool". Because of that, my boss perceives VDI as "cool". We can use that to bring in some of the other things we've been recommending for a while because its a fresh start. I do understand how it can be an uphill battle also though.


Couldn't agree more!

If you have the perfect tool to manage a physical desktop, you can easily manage a personal virtual desktop too.

If there are a lot more end-users than you have desktops, this might prove to be an expensive solution. That is why you want to be able to choose what type of desktop the end user needs, persistent, non-persistent, physical, virtual or mobile all managed by the same system including the applications!

That’s why we @ Tdist.com are very happy with the way we’re able to deliver a personalized desktop both physically and virtually including all the applications whenever they are needed with Scense (www.scense.com).


Extra points go to @Kevin-Goodman for the OPM buggy whip throwdown! www.youtube.com/watch


Great points about traditional VDI solutions, couldn't be closer to the truth.  However Unidesk solves all of these challenges.  Unidesk customers have 1 single gold image and deliver 99% of applications separate from the gold image.   Also our customers can support user-installed or departmental-installed unsupported apps, etc.  

Disclaimer:  I work for Unidesk


Well... generally I stay out of these discussions (since we're friends and I dont want to be seen as pimping my product and all that) but in this case I think there is a different philosophical argument being made that I want to get into a little.

The first problem is you assume (and lead the reader to assume) that some guy has been sitting around managing or trying to manage desktops for 20 years and now he has decided that VDI will fi it. This really isnt the case. In most case the folks running the desktops are early in their IT life. Maybe the Sr guys or managers aren't, but generally you find guys in their first years of IT (think year 5). Now this isnt the case in Goldman or some huge place with guys that have been doing desktops for 20 years, but you know what I am saying.

Also just because a set of desktops are managed poorly today (think about these colleges still using deepfreeze and sneaker net to do updates!) wont benefit from the mgmt of VDI (any VDI product). The reason these guys havent run out and put in SCCM or Altiris or all App-V in the last 5-10 years is because it IS complicated. You cant take a team that has done things "wrong" or really by hand for 5 years or more and expect them in 3-6 months roll out what is a really complicated SCCM rollout. App packaging, distribution, and all the pieces and parts that go to support a couple of thousand desktops broken down into groups of 20-30 desktops each that all need different configurations and different apps (maybe hundreds and hundreds of apps all in one business) .

Bottom line their desktop mgmt is bad... But to spend the time and money to get to a place where their skills cant take them or they cant manage long term, is impossible.

now let me step back in the way back machine. I remember a time when Brian and I and a whole lot of people ran around talking about terminal server and installing once for many. That message was understood. VDI isnt THAT DIFFERENT. And for some installing an app in a VM and cloning it off allows them the same benefits. You have to remember the great majority of IT folks out there are NOT App, or VDI, or SCCM or desktop experts. They are IT guys trying to hold a shop together.

Now my final thing is you and this Full clone thing. You tout full clones and 1:1 desktops. Which I understand. I like the 1:1 model except in maybe kiosk and lab places or transient type workforces... But with that said you say it and act as if that is the only valid model in your eyes... The "expert on all things desktop" ignores 1:1 models that still allow you Admin rights and full control of the desktop AND CENTRALIZED SHARED IMAGES AND MANAGEMENT!

You can get a 1:1 model with admin rights, that shares a gold image and supports all the way up to User installed apps right in the image. PVS w/ the personal vDisk (ringcube) does this today. Unidesk has done it since 2010.... Yet these models are somewhat ignored when you go on the full clone rant that most of your VDI articles run into. Why is that?

In that case you at lease wind up with the 1:1 model and control that users expect (as you noted) but still get the patch once distribute to everyone gold image model... with the storage reduction or the shared image.

long and short, SCCM, app packaging, app virt, etc is all not "easy". and there are a large % of shops out there that just Turn on WSUS and hope for the best, then package a few apps or sneaker net as needed. A bulk jump into that world isnt going to happen, just like Terminal Services isnt going to happen because while it has all the network connectivity knocks of VDI it also has the TS limitations and learning curve that we (those that have been around for 20 years) take for granted.

Sorry for this brain dump rant...


There are people in as yet undiscovered tribes in the heart of the Amazon jungle that can point to enterprises that after 30 years still have not got a handle on managing desktops.

a case in point  www.theregister.co.uk/.../dnschanger_rife_as_deadline_looms

"12% of fortune 500 companies and 4% of major US government organizations still infected with DNS changer."

That is, an embarrassment to the industry. But I challenge you to identify one production business environment that has implemented desktop virtualization that has this problem.




this article is complete garbage.

technology changes, requirements change, and businesses expand.

The world we live in today is not the same as it was 20 years ago, or even 5.

Someone very smart once said:

"Insanity is doing the same thing over and over again and expecting different results"

I personally like twisting that into:

"Insanity is doing the same thing over and over again and expecting to remain relevant."


@ lcelus.. I don't this so the article is garbage.

I've worked with different client mostly on the VDI projects and I think everything boils down to the applications and the flexibility for the users to use virtual desktops.

I’ve seen projects going down from XenDesktop to XenApp VDI.

I’ve encountered with bad ass applications which was very hard to virtualize with App-V.

But I think most of the time things got screwed up because lack of planning and lack of knowledge and some bad a** PM’s.

That’s why I like the Brian idea of having 2 projects and make sure your apps are virtualized before pushing virtual desktops.  


Interesting article.

One of the fundamental aspects of desktop design and service management is the adoption of build automation and that doesn't have to be too complicated or expensive.  

Automation either via MDT for smaller deployments or SCCM in the larger enterprise helps resolve one of the major issues that tactical VDI or broader scope projects introduce - fragmentation, fragmentation of design, fragmentation of architecture and fragmentation of operational management and  multiple builds for everything.

The adoption of automation at the heart of your desktop strategy regardless of the form factor will provide a transparent build process without the need for build docs, a consistent build process and a re-usable set of task sequences that can be utilised to build out desktops of various form factors and one that can be understood across skilled MS engineers.  Build automation will provide a level of portability of the desktop core service as a whole when looking to outsource, so benefits are there right up the service delivery management chain.

If you are lucky enough to get to the Citrix Synergy Geek Speak events or BriForum then book yourself in a Shawn Bass session.  Shawn provides a fantastic pragmatic and open approach to desktop virtualisation / transformation, one which mirrors Brian's thoughts providing business continuity or remote access through 1:1 or Citrix RemotePC initially and then start looking at transforming the whole desktop management piece as a whole later, otherwise the fiscal requirements to do both are likely to stop you from doing either.  You can't solve cultural issues with VDI.

A final point on the desktop virtualisation strategy as a whole:

Regardless of the scope of the strategy and whether it is to become your principal desktop delivery core service or not, there is always going to be fragmentation in design, architecture & operational management as you begin to leave behind traditional form factors that cannot / should not be converted and apps that cannot be served out of the data centre or virtualised. In any large Windows based enterprise this can't be ignored and it must be included in the program management phases and converged into design, build and service transition phases, don't have a Citrix team building out the VDI SOE and a totally different team building out the laptop and desktop SOE, instead converge them and unify architecture and management through enterprise platforms.  Citrix don't include this in the flexcast model  or in the desktop transformation accelerator (I know about client side virtualisation and I also know how tactical it can be and how difficult it is to transition into service and how it can fragment and dilute the desktop core even further).  Citrix however are doing a great job with SC2012 integration which is leading to design and management unity across all form factors.

Adopt and re-use build automation across all form factors, layer if need to converge your application delivery strategy and provide user experience continuity across form factors,  I agree with Brian - good desktop design, architecture and operational principals need to be embedded in the culture of the organisation before VDI is even considered as a magical way to solve all these problems, otherwise in the short term the service may seem simplified and easier to manage but in 12 months time it will have likely turned to ***.

Ref UniDesk and Personal vDisk - agree this can accelerate simplified shared vDisk management but maybe it doesn't solve the deep cultural issues that probably got them there in the first place? It is very innovative and has plenty of use cases but I feel it can complicate the architecture somewhat and reduces its portability through potential service providers as well as duplicating apps in the user space and possibly turning the user space into a potential management overhead, with that said I am a fan of both platforms and looking forward to future evolution - just being pragmatic.

When thinking VDI, think bigger - think about pulling xenapp, xendesktop, traditional laptop and desktop builds together - don't ignore the design what's left behind, design the core service to be adaptable, modular and manageable as a whole, invest in fundamental desktop design principals at the beginning especially around automation even before you get to PVS and VDI, do this and you'll be off to a winning start - however pretty darn hard and expensive to do retrospectively which is why most organisations that I have had exposure to are attempting to do this during their XP to Windows 7 migration programs where there is more budget to 'do things right this time around' and provide an architecture / service that caters for multiple form factors.