If VDI is so great, then why aren't you using it?

But I do use VDI....

Well a combination of VDI/RDS - full screen or seamless apps!

A good portion of th apps I use rely on a network/domain connectivity and I'm fortunate enough to always be near a good connection.

On top of apps a good portion of the data I work with is network/domain dependant also.

I try to never give our users something without feeling the "pain" myself. That way when the complain I can say "it works ok for me" :)

However since I've been doing a lot of RemoteFX testing I'm finding it hard to go back to traditional remote protocols!

Yes local horse power is going to be better and we still use it where it makes sense and I guess thats the point.... VDI/RDS/etc have there places for many use cases but not for all users.

but i do too...

my 1st device is my BYO laptop with strealed and published XenApp apps. i also have on my desk a Xenith thin client to a win7 vdi desktop i use as "always on production environment". remotly, i more oftently use published apps rather than full desktop...

it isntrue that i have problem to loose the "i have control" of my own laptop withnpersonnal use but i found a good balance the way i' working...

Thank you for saying this loud and clear. It's time people get it in their heads what is hype and what is reality.

but i do too ...

iPad for instant on / drop box file viewing, apps, battery life - wyse pocket cloud to Win 7 VM in Atlanta (mail, docs, dropbox, evernote, etc)  ProfileUnity™ (have my settings apply to three other machines i log into for demos )

MacBook - Open Client to VM [but i do use local mac apps a TON]

Physical Win7 Tower - great coffee cup holder [ has not been booted in months]

Chrome OS Netbook - [ pending ]

For me VDI solves problems, but not all of them.

Question becomes: Why aren't more people using VDI? Are the alternatives simply better as Brian is implying?

Is the Cloud a better model? Do people really want "offline" which is a fancy name for locally installed or synchronized apps.

Is the target audience "remote task workers" or actually a small percentage of power users and road warriors?

For those early adopters, please let this forum know about your adoption, experiences, and feedback.

Good point.

I try to use the best tool at hand. When that's VDI, I use it. When it's local application with LAN connection to a backend resource, I use that.

My yardstick is speed and complexity.

VDI will succeed by being a viable choice, not by being the only choice.

So, I think that you're right, being niche. I'm not sure about the 10% number, though.

VMWare player on a Win7 laptop for demo and testing purposes. Fusion on an old macbook for production work. iPad with a client to connect to TS at work to get files I forgot to bring home.

VDI helps me do my work but I work in IT (just like 99% of the other VDI users). As long as CAPEX remains as high as it is today for VDI implementations, I don't t expect large adoption rates.

I think it happens when there is an easy way to connect a smartphone to a monitor/keyboard.  Right now most smartphones are easily powerful enough to be a think client.  Why buy your sales force a smartphone and a laptop when you can just buy the phone and use Citrix Receiver?

VDI isn't great, Desktop Virtualization is great.

Unifying management of centralized and distributed workloads transparently which in turns extends the benefits of SBC to CBC.

I wish we could drop the VDI terminology for a more specific one called Server Hosted Virtual Desktops.

After all, in my own opinion I would consider VMWare Fusion a type of Virtual Desktop Infrastructure (VDI).

"Windows 7 installed locally on a laptop is really damn good"

Mac OS Snow Leopard is even better :-D

How can VDI beat the multitouch gestures ?

I use it, but as others have stated, I'm more like a hybrid user.  I use my MBP, iPad, iPhone, zero client, etc. based on where and what I'm consuming.  I will disagree with you in regards to hype, I'm working with several customers who are utilizing VDI for the right use cases and they're saving money.

I view this market like I did with server virtualization.  VDI today is being used for specific use cases that fit what the technology can accommodate.  As the technology matures, so will the use cases.  5 years ago, no one would have even considered running tier 1 applications on a hypervisor, but look where we are today.  Just my 2 cents.

I agree with a lot of what everyone has said.  First @Icelus has it right that server hosted desktops need to be called something other than VDI because all of the technologies on the markets whether it is local hypervisors (Type I or II), terminal services, blade pcs, or dedicated server hosted desktops they are all part of a desktop virtualization infrastructure.

Maybe I'm missing something, but I don't know anyone that actually works in this space that actually says what you called VDI is going to solve all customer use cases and is going to be the only desktop virtualization solution deployed to solve all use cases.  The only people that I see saying that are people who actually don't work in the desktop or virtualization field.  Hosted server desktops will be used in our environment to address specific business use cases and problems, but will only amount to maybe 20% of our overall desktop virtualization solutions in the long run.

I have said this many times before.

If you picture the future with the majority of client devices having low grade computing power then you're in the SHVD camp.

Conversly, if you picture the future with the majority of client devices having high grade computing power then you're in the CHVD camp.

If you picture the future with a diverse client device ecosystem comprised of low/high grade  computing resources, then you are in the mixture of SHVD and CHVD camp.

If you are using SHVD on client devices with high grade computing resources then you are pissing away your technology.

Per Moore's law, I vote CHVD will succeed expectations once management and the device ecosystem matures but SHVD will always have it's merit.

Also,

Neither myself or anyone in the federal agency that I work for uses VMware's coined definition of "VDI" (aka SHVD) because I am currently in the process of getting a PoC started for XenDesktop.

Maybe in a couple years I will be on XenClient with my XenDesktop and we can have a conversation then, but wait... My virtual machine running on a client hypervisor rather than a server hypervisor is not really apart of a Virtual Desktop Infrastructure so I guess I won't be using it then too.

Maybe when I just use my smartphone to hook it up to my monitor and peripherals then I will be using VDI.

Thanks for the clarification!

Don't get me wrong, there are certain places in the agency where we need to limit the client footprint extensively.

Fujitsu's Zero Clients look nice. An LED display with Power over Ethernet. It's going to be nice to see the client device market diversify.

Brian -

What is the problem with the following quotes:

 "I think there is a world market for maybe five computers."

 "640K ought to be enough for anybody."

 "Don't kid yourself: VDI is a niche. 10% max* at best. Mark my words. 'VDI will be 10% max. That might be 10% of all users, or 100% of users for 10% of their apps.'"

Only on of them can be unequivocally proven to have actually been stated so it will be interesting to see how this one plays out in the long run.  And your evidence for going out on such a limb?  Everywhere you go you see laptops.  Working for VMware might drive me to look at the world through rose colored glasses but I don't see how such a casual observation can be used to back up such a grand statement.  Having traveled to and from SFO enough over the years it's even possible you may have seen me sitting in the airport with my "laptop" too.  Of course, what isn't readily apparent is I don't travel with a traditional laptop but a Wyse Thinclient with stickers on it so no one really knows what I'm sitting in front of.  My desktop?  It's a hosted desktop back at corporate . . . the same one I spend 90% of my work related computing cycles in.  The other 10% of my time?  Well, I'm an iPad slappy and I'm not going to connect to a View desktop to check e-mail or look at a Youtube video when there are native apps for that.  I will when it comes to validate a Visio diagram though so I'm not completely decoupled from a Windows OS with my iPad yet.

You're other evidence?  You base it on the people you interview and communicate with on a daily basis.  I believe it's safe to say that the vast majority of these people are IT admins who, by the nature of their business, have the keys to the castle and means by which to bypass the same controls they place on their users.  This would be no different than my polling 100 people in downtown San Francisco on their thoughts around any Republican led tax initiative and then extrapolating that out as the pervasive thoughts in the rest of the country.  For every Appdetective, Glenda Canfield, Russel Wilkinson, Simon Crosby, <insert IT admin here> that reads this blog there are hundreds of thousands (if not millions) of users that don't even know any of us exist because they couldn't care less about IT the same way we do.

So to go and make a statement like "10% max" is quite a prediction.  But, let's say you're right.  Gartner says 1 billion PCs were in use in 2008 with a prediction of 2 billion in use by 2014.  Unfortunately I can't find an authoritative statement that indicates how many of these are used for corporate work versus those owned for personal use but let's start with the 1 billion and, as a conservative guess, say that 60% are home users that we'll never touch.  That leaves us with 400,000,000 PCs out there.  Using this as the starting point for 10%max that means we could expect to see 40,000,000 desktops virtualized.  I would be extremely selfish to think I could get all 40,000,000 so I'll split it 50/50 with Citrix (we can debate those numbers on a different day) and that leaves me with 20,000,000 hosted desktops or roughly $5 billion in licensing for us.  You know what?  That's a pretty sizable market that's worth going after.

I'm not a fool and I honestly don't expect to take down 20,000,000 desktops over the next year.  But the problem with any one of us using ourselves and our technology peers as a basis for making an industry perspective is we look at things myopically.  Let's look at the millions of other users that are sitting behind computer screens doing daily order taking, vehicle parts lookups, credit check scans, etc and then determine what the real percentage will be.

If you're ever curious, btw, of what it's like to spend most of your time in a desktop 2000+ miles away let me know.  I run a dual-monitor config connected to a Windows 7 OS hosted in corporate.  I still have fun when my non-techie friends come into my office and, after a few minutes of playing around on the desktop, don't have a clue that the desktop OS isn't sitting locally on my thinclient.

@Russel Wilkinson

Being able to sell the fact that a thin client can be visually indistinguishable from a fat client with a local install is one thing.

It's another thing to talk about the cost of having that desktop environment hosted in the datacentre versus hosted locally utilizing significantly cheaper resources.

If all the management and security benefits span from SBC to CBC, why force SHVD and thin clients?

Type 2 and hybrid hypervisors don't qualify.

www.ok-labs.com/.../vmware-mvp-how-it-works

OK, I had to finally sign up so I could respond to this one.  Almost three years ago, I was tasked with implementing VDI (server-based virtual desktops) for a client of mine.  After a brief shootout, we chose and implemented VMware View.  Today, we have about 500 virtual desktops.  We'll implement about 300 more over the next year and we will have completed our desktop PC replacement activity.

During the proof of concept, I put a thin client on my desk and turned in my client-provided laptop (my "other" laptop, since I already had my own Macbook).  I was very skeptical at the time, but would never go back to a physical PC.  Gone are the days of two laptops and an IPSec VPN for remote access.  No more 3-minute bootups in the morning.  I fire up my Wyse P20 and in about 15 seconds, I'm sitting at my XP desktop and all my applications are right where I left them the day before.  On Fridays (maybe), I reboot the VM on the way out the door.

Sure, updating a PowerPoint remotely requires a real good connection, but my 6Mb broadband connection takes care of that.

You couldn't pay me to go back to a Windows-based physical PC.  I use my Mac Pro at the house and my Macbook when out to get to my VM.  We have significantly reduced travel costs related to IT projects by providing temp contractors with rapidly deployed virtual desktops.

If VDI is the Matrix, I've swallowed the red pill - and completely happy with the results!  And it's only going to get better.

Two words: waste stream

while I use a VDI infrastructure at work it still is not a replacement for a full desktop, The performance is not there and the cost is still to high.  I think that the day that a service provider is able to host a desktop for me for $10/month then VDI will be in its prime.

Unfortunately....VDI sucks.  Should be renamed to TRDI - Training Room Desktop Infrastructure.

So if I can get all my apps running locally securely with a great user experience and managed cost effectively on my laptop great, I don't need a hosted desktop solution. Oh yeah, it's a Windows app world still in the enterprise, and nobody killed it off with Java or webapps. I need to provide access to those apps from many devices, how do I do that securely now and in the next few years for the vast majority of enterprise use cases?

Maybe that cloud thing will solve for this when connectivity will be everywhere and performance is better remotely. Even when that happens will it kill off laptops because I bet local will always be better than remote for many use cases.

That's the punch line, many use cases for many app types, means can't just think about VDI. It's more complicated than that. VDI will be a use case for some as will RDS for those who want it cheaper. If it's 10% of desktops then that's billions of dollars to go after and hence we will continue to see investment. However better more broad solutions that address more desktop/app use cases represents an even bigger opportunity. That's the trouble with our industry, we talk too much about VDI and act like sheep when we all know deep down the world is far more complex than that.

but I do too... well in a hybrid way like Matt said

When I am working from home I VPN in to work and launch my XenDesktop VM. The desktop is in the datacenter and runs apps really fast. I have easy access to data on the internal network and I don't have to worry about FQDN naming. I have couple of friends who have BYOC machine as an mac laptop and they do the same thing.

I agree with most of the points by Russel Wilkinson. Brian, you are forgetting that most of the people you meet are geeks and love to completely control their machines and environments. I can't remember the last time I called helpdesk to fix my  computer problem but most of the non-techie friends in my building love Xendesktop and how it just works which saves them time and frustration when they have to call helpdesk, so desktop virtualization is making an impact.

Totally agree with you brian...

...but I have to mention one thing:

All you said about VDI applies to TS/RDS as well.

I think VDI can be used in more cases than TS/RDS. Sure, today TS/RDS is cheaper than VDI but I think VDI may get rather close to TS/RDS in the future (changes in VDA licencing, improvments in storage technology or broader use of local storage and so on...).

Because you'll need to have a mixture with traditional FAT clients (about 80-90%!?) it will be easier for the people to mantain their apps for only one plattform instead of maintaining them on a client OS (FAT) and on a server OS (terminal server).

So I think the goal for VDI is to get as good, as easy and as cheap,  that it can replace TS/RDS.

But as I mentioned before:

The dream of a broad replacement of traditional FAT client infrastructures by VDI is over !

And why?

Because FAT client infrastructures are cheaper and easier to operate than VDI (AND TS/RDS !!) infrastructures !

Brian,  as others have hinted , I suspect you are too techy to use VDI full time.  You like the control and messing about that a PC gives you.  But you are a very specific user-type.

Personally, I think Windows 7 on a fast laptop is great, but the trouble is that Windows on a laptop doesn't stay great for long, and you also spend far too much time waiting for your offline world to catch up with the online world.

I travel around Europe, the Middle East and to the US but last January the frustration of the PC laptop just became too much.  Too long to boot, waiting for Outlook to sync, waiting while it was more interested in talking to itself than working for me.   The PC laptop has sat on a shelf ever since, and I wouldn't have it, or even a bright shiny new one, back.  It has been 100% cloud-based VDI since then.

After trying various approaches, the one that really works is a Windows 7 vm, hosted in the 3rd party cloud infrastructure that Wyse uses for most of its IT.

Access is from whatever I have to hand:  Desktop dual screen thin client in the office, Mac with an RDP client at home, iPad or Android smartphone with PocketCloud, or a Wyse thin laptop on the move. (more details on blog - see bio for link)

Even when I broke it (still don't know how but even our IT guys couldn't make it respond), they provisioned me a new one in minutes.  And in those few minutes I used XenApp to access my data and continue working.

The simplicity of the approach is what I like.  I can get to my desktop in less than  30 seconds from any device.  It's always exactly how I left it and in the rare cases I need something local, Dropbox is a great tool.

Less than 10%?     We are still early in this market but I'm convinced carrying your processing power with you will become a minority sport.

@Brian

The advantages that people are talking about for SHVD versus Local are based off of the perception that SBC provides a better experience than CBC.

But what everyone is missing is that the only reason is due to management or lack thereof.

If Client Based Computing was managed the same way as Server Based Computing, what are the reasons for running your workloads in the datacenter?

Is it for cost?

Is it for security?

Is it for management?

Is it for user experience? performance? portability?

Computation is computation, doesn't matter where it is.

If SBC management is extended to CBC, the needs for SBC execution get limited.

The only reason SHVD are all the talk is because we currently don't have a mature CHVD solution.

In 2 years we will all be having a different conversation.

Some people are starting to realize that we can extend the "Secure By Design" datacenter architecture to the end point.

Gone are the days of the past 20 years that IT classified SBC as being the only thing that is "Secure By Design".

@Brian   "I just can't honestly believe that anyone would WANT to depend on a remote desktop when traveling. To me that's the best case for LOCAL dekstops!"

You'd think so wouldn't you; and pushing VDI beyond where most people would think it is natural was partly in my mind when I ditched the PC.  BUT, having being cloud-based for a year now, I prefer this way of working. I just find it simpler, delivers more consistent functionality, and I spend less time being my own IT manager - which I was with a PC laptop.

So why aren't the customers and vendors you sit in front of using it - good question. I'd hazard a guess they are more like the frequently mobile user that you and I represent; and so not the easiest, most obvious place to start with a VDI project.

Certainly the volume of thin and zero clients we at Wyse see being shipped and connected to hosted virtual desktops is evidence of strongly growing adoption.

Your point on the processing power in mobile devices is well taken, but my minority sport comment was targeted at the power used for the heavy lifting of applications.  Using local processing power helps with user experience and overall system scalability, but my view is this power will increasingly be used for client-assist or collaborative processing, not running the main app itself.

And fast-forward a few years,  will that app even be Windows-based?

@brian, even if I only use office, my data is not secure on the laptop unless that laptop is full disk encrypted with tight policy, or I use a hypervisor approach. That's not what I want but it's a requirement because of the apps. I agree the local user experience is better but it's about data security for me. Things like XenVault are good concepts that may make this achievable in the off-line mode for certain apps. Sybase etc also try to do this, as well as emerging mobile device management companies like MobileIron. In my case I use a laptop most of the time for personal stuff, and just connect to a remote desktop/app to do all my corporate app stuff. It gives me the flexibility I want and the ability to provide security that I need. I don't really want to be the user device management business. The cloud based services are not even close to providing this level of security. For example there is not way I can put certain files in dropbox, that sapce is just not here. So  it will be mix of solutions for different use cases, 1 of which you describe for a certain app sub set.

@Mub, totally disagree with you that managing RDS/XA is more expensive than PC. That's just plain wrong and just confirms for me that there probably many other's our there who think the same think and also confuse this with VDI which is cost neutral at best today vs. a PC.

@appdetective, I don't confuse RDS/XA with VDI.

I've done a lot of ROI/TCO calculations in the last years and unfortunatelly (I'm not happy about that!) you can see, that in most cases a well managed FAT client infrastructure is cheaper (looking at OPEX) than RDS/XA AND VDI.

These calculations are quite complex, but let me just pick some points out to show you, that it is very hard to get a real ROI from this technologies:

1.) You still need a FAT client infrastructure (e.g. for your laptops) => because of that you have to maintain and support two different types of infrastructures instead of one

2.) If you look at some outsourcing offers on the market you will see, that a RDS/XA Client isn't  really cheaper than a FAT client (and I have some kind of insight in that calculations).

3.) The main savings (for RDS/XA/VDI) came from using thin clients. But most of the time companies are using a - let's call it "FAT-Thinclient" with XPe or windows embedded OS installed. That's because some applications still have to be executed locally (...and with more an more companies using VoIP I think this will still be true in the future). If you are not able to get rid of your deskside support you are loosing one of the biggest saving potential for this technologies. And (again unfortunatelly) in most cases you still need this deskside support in any way.

4.) To save OPEX you normally try to "left-shift" the helpdesk costs today; that means that you try to solve as many prolems as possible by the first level support or even by the user himself (e.g. passwort self service reset). With RDS/XA you "right-shift" your service costs to more problems to be solved by high skilled people (people who are trained for RDS/XA). Perhaps with RDS/XA even a little bit more than with VDI, because you can still use traditional remote support tools on the vitrual clients and the first level still has access.

To make a long story short: You can only save money by introducing a new technology into an existing infrastructure if you are able to:

1.) reduce the number of operational tasks

or

2.) reduce the time such a task will take

There is no other valid calculation.

Be true to yourself and ask you: which tasks I don't need anymore and which tasks will be done faster (because of using RDS/XA or VDI)....and after that look at the tasks that may took even longer and that will be added to your operational work by introducing these new technologies...and perhaps after that keep in mind, that you have to compensate the project costs for introducing these new technologies.

Don't get me wrong...I like these technologies and they can provide essential benefits but I don't think, that in the most cases you will end up in reduced operational costs...not with VDI and not with RDS/XA.

I know that the vendors have done a good job to convice people to think: "RDS/XA is cheaper because it's central", but my experience is a different one....and keep in mind Brian's great article about cost models (www.brianmadden.com/.../how-to-lie-with-cost-models.aspx).

Perhaps you (or your customers?) were able to realize a positive ROI in a acceptable time, but I think that is not the normal case.

OpEx.

The cost with PCs is the amount of patches and updates that have to applied to a distributed infrastructure. This has to be applied to many machines by many isolated groups which means it's a nightmare to control. This results in poor service and then I have to run around with inventory and so on to validate that everything get's there. Every patch Tuesday is therefore a PITA and every zero day patch is like farting in the wind and hoping the wind blows away from you. It's just too slow, not robust enough and can catch me with my pants down.

Branch office and Network costs

In my case I have built remote offices using both RDS/XA and VDI. In the case of RDS/XA it's very simple. I get to remove one my largest costs, application management. One change and all the users get it fast (no need for all that App virt BS either which does not always work), I also don't need to put heavy application infrastructure locally and that saves me a ton of network costs, although the occasional spike can still impact me and QoS is not really possible with the remote protocols today.

Cheaper Application infrastructure

Following on from the previous point, avoiding fat local infrastructures, means cheaper data center space, thinner buildings, cheaper staff etc. It also allows me to have more chatty apps that pump more data directly to the app or desktop in the data center. No need to compromise on network and instead focus the app on getting the user more of the stuff he/she needs. I also avoid duplicating application middle-tiers around the world that cost a lost of money.

Labor costs.

As I move to less data centers I can centralize my support staff to cheaper parts of the world. That's both data center staff and user facing staff. As there are also less moving parts especially when it comes to apps, my testing times are drastically reduced and I run a plant with much greater ratio of admin:desktop. Prior it was like 1 admin to every 100 desktops or so, and now I can drive that to 1000+ with some automation that I have worked in addition to the vendor stuff.

Hardware

I 100% agree with you on the client still needing some local execution in many cases. Sure there are people who treat Wyse with custom $hitty OS as religion and think it's more secure and then *** about client fidelity, they are just dumb. Yes there are valid use cases for thin clients where everything can be executed remotely 100% of the time and you son't give a S about VOIP etc. However, that does not mean your client costs are still there. Firstly you can depreciate your existing fleet over a longer period of time as it will run less local apps. You can also deploy a thinner WIndows build that is essentially one image and just deliver that as needed. Much simpler and lower impact on labor and skill set. In the data center your hardware costs are higher than a PC, but those get blown away with RDS due to the density especially with x64 plus all the other benefits.

Mobility.

I now also have an option to not provide laptops for the vast majority of my users whose experience sucks on them anyway due to the security requirements and resultant software stack which increases my costs. I can do things like BYOC and where I must provide the device I can offer pools that are rented and upon return have my teams rebuild them to take away the burden of remote patching and management. Type 2 solution like Moka 5 give me more options to deliver offline use cases and I can still enable users to get to more important stuff that requires an online connection due to the application backend via RDS/XA. Then there is this tablet and smart phone thing. There is no way I can manage all those devices unless I start to build more BES like infrastructures, invest in companies like MobilIron for native apps which there are few. So now I have an option to connect back in from many devices and my users have something they want to try, what a change :-)

So respectfully I totally disagree with you that the traditional way (even if well managed) is better. It's is simply not agile enough, and is the wrong architecture to bet on for the future especially if one is about to spend all this money on Windows 7 to add no new capabilities that enable users. Now that does not mean new centralized management approaches geared towards client side devices are not a valid alternative. I just don't think they are mature enough to matter today for the vast majority of people. So yes it depends on use case, but I don't buy your argument that the vast majority of customers who use RDS/XA and understand what they are doing end up with a higher cost. That's not a vendor statement, it's fact that's been proven time and time again. When I stick to PC's it's because the experience is better for the user for certain apps, but I address that as an exception. I need to balance that with cost of hardware, mobility, user satisfaction, security, speed to management, labor costs etc. Sticking with FAT PC's with traditional management tools is simple not going to cut it. I will up in a world where I will offer hosted and locally executed solutions, but they won't be managed as distribute services. They will  be managed as centralized resource because that makes more sense and is where the world is going.

@mub

It doesn't matter how well managed a Traditional PC is, it will always suffer from the syndrome of re-inventing the wheel.

ex:

Having to install the same application for multiple users, or having to patch multiple workstations.

Hardware, OS, Applications, etc. all introduce potential problems, and when combined can become a cluster **** to support over time. They all have their dependencies.

All of the technologies you listed are just fancy names given to tools that allow you to perform abstraction of these necessary components so they can be managed individually.

If the technologies are usable within your application/desktop infrastructure the cost savings are inherent and if I can replace the need of 2 tier-1 helpdesk staff with 1 tier-2 support analyst then so be it.

Good question.  I am using VDI.  I have a PCoIP zero client with dual display at my desk and connect to a Win7 desktop on VMware View 4.5.  I access this desktop remotely from customer sites around the world.  

Teradici has zero clients in all meeting rooms so it is easy to connect to my VM from any meeting room. When I am the presenter, I don't need my laptop mobile client) as I just use the zero client in the room.

I have a PCoIP zero client at home and connect through Teradici's new Security Gateway (which is being integrated into VMware View) so no more need for VPN, just a single click to securely access my VM.  This could be a View client on my home PC, but using the PCoIP zero client for security - and to eat our own dogfood (mmmmm).  

I have a MAC which I bought prior to VMware View supporting PCoIP, and I use it mostly as a mobile View client (VMware View 4.5 client in Fusion).  I often use this during meetings where I need to attend, but I am not driving the meeting.  

To my wife's dismay, I geek-out with other IT guys while watching my kids in Taikwondo class as I connect back over Wireless & the Internet to the office and crank through my Win7 desktop.  They are blown away with the performance.  

Mac-wise, I use the MAC for my itunes and video editing (some biz, some personal) with iMovie.  I have MGI Photo/Video suite on my VM, but I like the imovie controls better.  

As the Director of Systems Engineering at Teradici I am seeing a lot of growth in VDI - on the early side of the new technology adoption curve for sure.  Crossing the Chasm by Geoff Moore is a good read about the adoption of new technologies like VDI.

@stuart since you want to use this as an infomercial let's also have some disclosure.

blogs.virtualizationadmin.com/.../vdi-remoting-protocol-comparison

1) Not too many smiley faces when you require somebody to buy a custom hardware acclerated thin client, which is complete $h1t to shove down somebodies throat and lock them in.

2) Your stuff sucks over the WAN even with hardware acceleration. F'd my test network I can tell you for sure.

3) Be clear that the software version of PCoIP is a POS compared to the hardware accelerated version.

4) Complete BS video below that I am sure you will be using in your marketing that is 100% lacking details to support your claim of apples to apples.

www.youtube.com/watch

5) Why is it that you guys and VMware continue to lie about what you can really achieve in the real world with PO$h1tOverIP, make BS comparisons, shy away from independent bench marks and tell all the people at VMWorld who know the truth to F off when they try to present their findings?

6) When will VMware publish what % of their View use base uses RDP vs. POSoIP. I bet you they won't because the truth is they are all on RDP. All 10 customers with 50 user pilots still trying to figure out why the F they are spending so much on just VDI and a soaring vSphere bill.

@appdetective

Nice reply and an interesting discussion.

I'd like to answer to some points you mentioned:

First of all I totally agree with you, that centralized infrastructures (like RDS/XA/VDI) have big advantages when it comes to flexibility and dynamic.

But if these advantages have an major impact on your OPEX is another question.

So, NO I don't think that "the traditional way" is better, but YES I think today it is cheaper in most cases.

So I'd like to reply to some points from your posting.

1.) Patching is faster and can be rolled back easier if needed. But you still need to test all your applications with these patches. The distribution of the patch is a single click with a traditional software distribution as it is with RDS/XA/VDI...the time it takes the software distribution solution to install the patch on every client has no impact on the patching costs (but again I agree that it is a good thing to have patches deployed quickly and to be able to roll back again).

2.) A valid point when it comes to costs is the integration of branch offices. Indeed if I'm able to use a centralized infrastructure I will save money on supporting decentralized servers and (perhaps) I can also save WAN costs or even consolidate my data centers (if I have that much today). But again, I still need some kind of local execution (e.g. VoiP, some incompatible apps, ...). And if I need some more demanding things like multimedia and USB redirection, I'll end up quickly with decentral RDS/XA/VDI farms and my savings are less than expected.

3.) I really don't believe you can save any time for application testing by using RDS/XA, but may be we have different customers/experiences. But OK, let's assume you have less effort, why don't you use application virtualization on FAT clients (providing the same benefits of single app management and isolation to the OS) and save your money for a very complex and expensive central infrastructure for RDS/XA?

OK, the answer is clear: With vApps on a FAT client you still have the traffic and latency for your local apps to the datacenter and the other way around. But it is only a valid cost argument if you really need that (so only in WAN environments) and if you can realize it (see point 2.).

4.) When it comes to the supporting costs of a "FAT client" vs. a "FAT client with an easy and standard image", ask the outsourcing or service partner that you trust how much $ less you have to pay.

I think (and know from many customers) there will be no differnce. But sure, you may be able to use your hardware for a longer period of time and save some money (OK, I've seen some customers extending their PC lifecycle to 6 years with windows XP but I think that's not usual) but what about the normal refresh cycle for your central server and storage infrastructure...I think it's about 3-4 years...so don't forget the additional lifecycle costs in the datacenter, eating up your savings.

5.) Totally agree with the benefits for integration of new endpoint device (like iPads, ...), as it is a benefit coming from the bigger flexibility. It can also have a cost impact if your company want's to provide iPads to their employees, but if that's not the case you don't save any OPEX costs.

6.) For mobile endpoint devices I was very impressed when I took a closer look at the Mirage product from Wanova. I think their concept of supporting mobile devices is a great approach (and NO I'm in no way involved to them) and I can only hope that a big vendor will invest in them so this idea get's the attention it deserves (don't want to get into details here; Brian made some good postings about this). This idea may be integrated into some client hypervisor products and then you have a great and cheap technology, that provides many of the benefits and savings you mentioned). But of course that is "client based VDI". So - because of the costs - I agree with Brians prediction (www.brianmadden.com/.../prediction-90-of-the-future-quot-vdi-quot-will-be-client-based.aspx)  

Don't get me wrong, the points you mentioned are all valid and may provide some savings but these savings have to compensate the cost for learning and introducing these technologies and the additional operational costs (RDS/XA/VDI doesn't come for free when it comes to OPEX).

If you are able to

...reduce your costs for onsite services and hardware refresh through thin clients

...reduce WAN and branch office server costs

...don't have to support two different infratructures (only some remaining FAT clients and not a 30:70 ratio or even worse)

then you may be able to see a positive ROI...even if it is a small one.

But how many companies are able to achiev this?

I totally agree that if your target as a company is to get more flexible, dynamic and may be scure this is the right way for you.

And it may be that your company can react more quickly to the market and can make more money in their core business.

But I don't think you should do it because you you can save a lot of money on your OPEX...and that's true for VDI and - I'm sorry - in most cases for RDS/XA too.

@Stuart are you for real?

Using this forum to sell used cars

you should be ashamed

@mub yes good discussion but I'll keep this response as I need a break :-)

Disagree on ESD with you, Time is the cost as well as accuracy and total lack of agility. Distributed computing is just not the way forward. The testing is equal in all cases since you have to test your apps. In fact with VDI it's the same OS so no extra cost. As a result App mgmt remains a huge cost of the PC which is all about Opex reduced by other model if done right. lot's of stuff out there to back that from Gartner etc.

You can deal with local execution in a number of ways that don't end in fat client support staffs that increase OpEx. This is why I think Reverse Seamless is so important as a use case to enable these use cases.

The extra testing on RDS/XA for the most part if a fad. 9/10 a simple test will let you know it works in a multi user mode. Very little time added to the package lifecycle since it's so common. Also again with the VDI model no extra time is added since it's the same OS. Also a lot of the problem goes away with R2 and Windows 7 thanks to MS making them the same kernel which was a cause of a lot of pain in the past.

The example of a single fat client image for all is a niche use case, where it may make sense to stay fat, if you ignore all other capabilities . Real desktop users are very different and hence the need for better personalization.

The refresh cycle in the data center is lot slower than the PC and with virtualization the usage of those parts is being driven up. If I can get dozens of users on a machines with increasing density the argument of do it cheaper on local looses $$ value very quickly minus all the benefits of running stuff locally for user experience.

Agree with you on Wanova as an interesting concept. They also don't require a hypervisor which gives them an edge over Unidesk I think without having spent enough time thinking about it yet. That goes back to the central management point with local execution, and I tend to agree that remote hosted when looking broadly

will go that way and solve the problem. However many will go all central (cloud thinking)  and all vendors pushing it including MS, so the central mgmt  local execution in the desktop world may take a long time to get any real traction. Also there are also the usual security concerns and all the previous stuff I talked about which will drive this to a laptop model only IMO.

Of course, we at Quest *do* use our own product internally. See www.quest.com/.../QuestonQuest_VDI_CaseStudy2.pdf for more details