What the Windows desktop will look like in 2015: Brian's vision of the future

Hmm.. Here is my 2c, think about Windows XP mode. In future there is something like Windows Mode (TS,VDI). You will run you apps in "cloud", but if you need legacy Windows Apps you start Windows mode. MS, Google etc. are moving their apps from desktop to "cloud", running Windows on your computer is not mandatory anymore.

That's a very interesting take on how the desktop might become entirely cloud enabled. As you explain it, it seems obvious, and only an extension of where we are going already with VDI. This is not to detract from your vision, only a complement of how correct it feels.

Initially my reaction was that large companies - those involved in secure transactions or with high security goals would be unable to embrace it.

However by explaining that the "cloud" could be privately managed with only ISV apps and OS's being fed in from outside this is a workable model for all.

One further thought is that with the arrival of netbooks and iPads there is space (which Microsoft might fill) for a lighter OS to become the de-facto standard for offline/online use, I feel win7 is too large and contains far more than most people need, Perhaps win7 could be made more granular giving users the choice of only the features they need.

Brian,

see my answer here: blogs.pfuetzner.de/matthias

Matthias

Brian, I enjoyed your article. A few thoughts.

You correctly observe 16 bit apps are still around. Frankly I think this is holding back the adoption of x64 in many organizations, thanks developers of the world! IMO I think even with the continuing maturity of application virtualization we are going to be in a mixed world of traditional and virtualized apps for a long time. Not all developers are going to buy into infrastructure deployment best practices overnight. So while I hope we will be a long way forward by 2015, I think history teaches us that it will take longer :-(

As for layer cake, I think there will have to be many hurdles crossed just like in the early days of TS. App compat, management, 3rd party support immediately spring to mind. Layers sounds good, but with it comes new complexity if not done right, what Windows can actually do and of course importantly customer acceptance of a new model. I’ve heard many people say basic personalization is good enough along side existing systems management tools. They don’t want layers imposed on them. What ever the end result is I hope it doesn’t take 20 years to mature and it’s great to see some really smart people trying. If XP is EOL in 2014, which will catch many of the stragglers to move to Windows 7, Windows 8 likely in my opinion will only likely start to shift in this direction if at all in combination with IT services on demand. The one after that, I have much more hope for the mass market. Of course that does not mean smart startups won’t innovate which is always good for the industry.

In terms of delivery, you touch upon a trend that I very much agree with. The world is moving towards a service delivery model. I don’t think in the future people will need to always buy perpetual software licenses, pay maintenance, negotiate enterprise agreements etc. Consumption based services lend themselves to more customer flexibility. If I don’t like it, turn it off and use an alternative. Does IT need to own this process? This brings up an interesting paradox for many, how much customization should you be putting into your IT systems vs. the ability to choose off the shelf? What is the future role of IT? I think this is a broader consideration than just desktop virtualization.

So sure the future desktop could be a mash up of various services that are consumed on a rental basis. This is why user based licensing to me makes so much sense, it enables these models. However to adopt this, a lot has to change has in ones organization. A transformation process has to take place to get ready to consume these services. Distributed computing makes this difficult, and therefore I believe that desktop virtualization is really an architecture that is part of a broader strategy to consume IT as a Service.

Average windows admin is an idiot, blindly follows App-V as the only solution, locks into SA/MDOP, Q: hard coded mappings, lack of IE virtualization etc. There was even an article on DABCC.com that I saw a two weeks or so ago, that points to MS recommending Med-V and TS for IE 6 etc. What a joke. The only guys that can do it is Xenocode. VMware lie that they can, it doesn’t work, fud machine at it’s best. Citrix’s Joe Nord is trying, tick tick. Xenocode also kicks the crap out of any of them in streaming anywhere period. So with crappy virtualization technology, I agree it will be mixed world for a very long time.

BTW Atlantis is not a layering solution anymore. I’ve always said they were crap at layers. They are rebranding themselves as a storage I/O appliance. Their only customer that they keep touting is a large financial that has a few thousand users, planning to expand.  Hopefully they succeed at that, because the layering solution was crap. I agree layers has a ton of potential, but young, very young and immature with no supporting eco system. Now add MS back into the mix, and you are dealing with a bunch of dinosaurs that just keep making $hitty Systems Center even more complex. How much crap are they going to keep bolting on. I like their new management service offering. Perhaps somebody with a brain has woken up at MS. SCCM in the cloud in the future.... Then I can turn the crap off easily and explain to MS why it can’t actually do anything fast enough and go elsewhere.

Agree with patching problem. Pain in the butt that needs to be solved. It’s a stupid problem that still exists. Somebody should just solve that with VHD delta’s. How hard can it be?

Personalization in the cloud is a good concept. Good area for Appsense/RES to go for. Today both these vendors are too complex for just profiles and desktops for the vast majority of people IMO. The 40% of Citrix customers in Europe us Appsense claim in an earlier article is just BS. I would love to have Appsense have Citrix verify that. It’s marketing at it’s best.

I agree with all the thoughts, about desktop virt being transformational and a step function to consuming services. This is why just serving up apps makes much more sense than VDI in many use cases. This is certainly a future direction. I hate the big systems like SCCM that miss features so you end up getting others in the mix at scale and don’t use half the features of the big system. Crappy tool sets like MDOP that are used by slimly stupid MS sales teams to sell vaporware. I mean come on, a while ago MS was trying to tell me that Med-V is a great idea for app compat. No it’s a stupid idea, I can use TS for that cheaply. If I need offline, give me Type 1. Anything but Med-V. Hey I may even consider using crappy app-v if I can actually figure out with my global dev teams which hard coded drive letter they are going to remove from their code base and never use again. Yeah right, Crap-V.

So with so much fud, bad technology and dumb people in technology, 2015 I don’t think. 2020 is more like it.

Brian - Great article (although you really should have learned about putting dates to these things).  A couple of thoughts to add...

Installation can be considered to consist of delivery of the app, ensuring dependent pieces are there, configuring, and licensing.  Licensing can, and will need to be, moved from installation to run time for all of this layering to really work.  This is possible, but we still have some issues to solve for "offline use" scenarios when installation goes away.  Solvable, we are not there yet, but close enough that the end of installation could happen in the future.

I still see the data as the bigger issue in the short run.  We don't talk about it much, because we don't really know what to do about it.  It is not as simple as "what to keep and what to through away", although understanding that would be a major step forward.  In my 2008 Briforum session ("The Data Problem"), I called out eight basic categories of data that a user generates and these need to be separated and treated differently (I'd post a chart from that session here but I can't do that in a comment).  To successfully break the bindings of one user = one OS image in a WAN world (not a server farm where your bloated roaming profile is just a SAN partition away) this will need to be addressed.  Quite frankly, (and especially if we continue to live in a windows world) I think we need Microsoft to take the lead in solving the data problem.  Not only do they control the platform, but they provide a heavy hand in how all of these applications are developed (this has been a key to Microsoft domination: own the developers-->own the apps-->own the platform).  Unfortunately, I only see Microsoft as likely to try to address this in the form of new platforms for new apps (read: Azure).

We might not now how the pieces will work or exactly how we will use it all, but it sure is going to be a fun ride getting there!

16-bit Apps? What's the problem? It's not the apps, it's the OS, that's not capable of running those transparently. Take, for example, Solaris: With Solaris, there's no problem running 32-bit apps on a 64-bit kernel, as long, as you've got the necessary libs. Could have been extended also to 16bit, but, that's not been used in Solaris...

Take, for example, VirtualBox as a type2 hypervisor: As long as your underlying hardware is 64-bit (and has the virtualisation-support), VirtualBox allows to run 64-bit guests even on 32-bit OSes. So, where's the problem?

So, from that we can easily see, that the capability to RUN 16-bit Apps is a PROBLEM of the OS, not of the old legacy 16-bit app itself...

Stating this to be a problem is a limitation of the thinking, which is imposed by the limited thinking of the designers of the dominating OS...

Matthias

Great point Matthias but you also give way too much credit to idiot developers who don't know how to write applications for Windows. The are more idiot developers on this planet than there are infrastructure engineers who have to deal with all their crappy code. Yes Windows sucks in many ways, well then don't write for the OS, or get smart enough to write decent software that doesn't cripple your organization, or move to a new platform. That's real life, anything else is naive, delusion in a Windows world. Sun is wonderful. yeah right look where it got them. Show me one developer who loves Java IDE's. MS knows a thing or two on how to win developer hearts and minds, it's a shame most of them are stupid when it comes to how to write enterprise class software.

The reality IMO is that MS is like the Roman Empire just after it's peak. On it's way down. The OS itself is not worth anything. There is no reason to pay for an OS, or spend expensive cycles upgrading. Windows 8 is the next Vista waiting to happen. Who will upgrade to Windows 8 after they barely get to Windows 7 by 2014 due to MS forcing them via EOL of XP and locking their dev tools to a platform.

That's the problem, locking dev tools to a platform. It holds the world back, MS kills innovations with this. Azure is just legacy MS thinking rebranded as cloud.Why does anybody in their right mind want to lock into Azure? Why not build a component architecture that allows you the flexibility to build and consume from best of breed. This is why I like Amazon. They are simply a resource provider. For the the most part they do not impose their bias on me on what OS to choose, Dev stack, apps to use etc. I can simple use them as a resource provider. MS on the other hand like they have done through their history will provide me with half assed products that lack features and will require me to build on top or use third parties in Azure. Sure I may use some services, but to assume I want to marry Azure just because I like .NET is crazy. The world has changed, and that kind of legacy thinking will drive developers into the arms of competitors, like Google and Apple who are building armies of developers of their own. Yes they want to lock you in also, just like VMware does with VSphere but Google and Apple are having far more success. I think that is because they have OS platforms that are purpose built for function. iPhone, iPad, Android do what they do well. No heavy baggage like Win 7 with feature overload. That's why Courier is going down the toilet and why also vSphere is a POS, of course looks who's leading that toy shop, another former MS guy that wants you to lock into a feature rich platform.

MS needs a cultural change. Get over Windows, and start to innovate in a more modular fashion so people "Want" to consume their software and don't feel forced. MS has a lot of good stuff that I would consume if it was consumable without all the baggage and lock in. I think there is more money to be made doing that. If they don't all they are doing is empowering their competitors. It's time for fresh thinking and leadership at MS, be open if you want to win in the cloud, not closed. Azure to date is CLOSED and a dumb strategy.

Hi Brian,

Interesting piece and thanks for the mention. We’re already seeing the crossover of our solutions into the cloud. As the cloud becomes more pervasive so will the need to for organizations to deliver uninterrupted workspace access. The nature of cloud will open up more entry points and pose challenges in delivering the personalized settings users need, regardless of where they try to access their information. As we look forward, personalization from a user-centric perspective will become critical, taking into account not only what IT services users need, but HOW they use them. There will also be tremendous benefits on the security side.

Specifically in regards to the cloud and user settings: As long as we are looking at +150ms latency when accessing Azure or other cloud based services, I believe that a hybrid approach to storing the content of the user workspace is the way forward. In the case of user workspace configuration data (drives, printers, reg settings, etc) you're fine if you cache these items locally, hence the central datastore can have as much latency as it likes.

All things being said Brian, I believe you're spot on with the above breakdown of the technology stack. Understanding the impact of managing the user's workspace is becoming more important than ever.

Thanks,

Max Ranzau

Sr. Consulting Engineer @ RES Software

Good article Brian.

The way I look at this is that we now have many different ways to deliver applications and that we can choose the most appropriate technology based on the application and the user. But essentially, as long as we can get the application to the user we are good. Beyond that it is all about tailoring the application for the user and personalizing it.

Tailoring the application takes care of the business specific configuration that happens today when an organization repackages the application. It includes things like language packs to take account of where the user is and configuration changes for different part of the business. Personalization takes care of user introduced confoguration changes.

And to Harry's point on the long term role of IT. I can see a large part of that role becomes this customization of applications (both tailoring and personalization) however they are delivered.

Incidentally, I see both tailoring and personalization as being aspects of the user environment.

Martin Ingram (AppSense)

this is what i have been talking about for a couple of years - here is a live example: www.microsoft.com/.../6370b721-60ee-4a14-b7e0-73471397004a