“Help me manage users and then the devices will take care of themselves.” What does VMware mean?

"Help me manage users and then the devices will take care of themselves." This is a quote from VMware's COO Tod Nielsen during the Q&A portion of the company's 2Q 2010 earnings call.

“Help me manage users and then the devices will take care of themselves.” This is a quote from VMware's COO Tod Nielsen during the Q&A portion of the company's 2Q 2010 earnings call. (Thanks to Alessandro @virtualization.info for the great coverage.)

Continuing yesterday's conversation about the near-term opportunities for VDI to transform desktop delivery in the next few years, what do you think Tod's answer means?

I've written quite a bit about how desktop virtualization is different than server virtualization, and we heard Paul Maritz talk about the shift from devices to users way back at his first VMworld in 2008. So this "users not devices" message has been around for a few years. But do the products align with the message? Does View really focus on users?

Is VMware trying to spin their RTO Software purchase (to be called "Persona") as helping their user-centric focus? (To me it seems like RTO is a more tactical Windows user profile fix.) And VMware also confirmed on the Q2 call that Persona wouldn't even be in View 4.5.

Maybe VMware is trying to spin their decision to delay and/or cancel their client hypervisor as following the customer requests for more "user centric" solutions since client hypervisors are, umm... too device specific, or something? So a Type 2 client can be deployed to a user with device independence?

Maybe now that ThinApp can be licensed per user and apps can be deployed to groups, that's their idea of user management?

What do you think? When VMware says that they are going to help users manage users instead of devices, what does that mean? Is that something they're doing now, or something we're going to see from them in the future?

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

It is cryptic BS that really means virtualization.info/.../vmware-still-very-skeptic-about-the-maturity-of-vdi.html

I can't stand these F'ing con artists. Trained at BS kings MS and now trying to con the world to lock into vSphere. That's the truth of all this. It's all about vSphere, that's where the money is.

They really have no clue what to do, understand that VDI is making no money for them. Quarter after quarter soften the message and let the idiot analysts who don't get it slowly change focus. I just read the transcript this morning, and they are also not breaking out desktop revenue, so that just shows you that VDI is next to zero. Now I am sure Citrix will hide VDI under the XD umbrella when in reality people are buying it for XA but the idiot analysts will fall for it I bet and Citrix will simply say this is all Desktop Virtualization which is more than VDI. True, but then show the world the truth on VDI as well. Actually how do Citrix count, with a XD license.......?

Also their comments on View offline being a revelation because customers are telling them they are not ready. F'ing liars and any analyst who believes that is an idiot. No VMWare failed at Type 1. The virtualization kings had their a$$ handed to them. It goes against their datacenter view of the world and they well on their way of pulling out of VDI since it makes zero money for them. It's nothing but a free feature to lock stupid customers into ESX enterprise agreements.

Unidesk seems to kissing a lot of VMware a$$ these days, so perhaps they want buy them and tell us layers that are user centric managed is the wave of the future. It will only take another 3-5 years to get working, and by then who cares, you'll all be on Zimra email with all your crappy windows apps converted to Springsource running on a vSphere cloud that will automagically provision users and scale elastically. Of course you'll also want to lock into crappy ChromeOS and use crappy Google apps and you will be no better off that you are today. That's the VMware of the future, nothing to do with real desktops use cases. As far as I am concerned the number 2 player is Quest in terms of technology, but they have zero market presence.


I think it means that VMWare wants to recoup some of the $ that they are giving up to AppSense for managing profiles and desktops rather than devices.  

Also, now that Citrix has introduced XenServer, a lot of organizations are looking at an end-to-end Citrix solution and are moving away from ESX.  I think that VMWare is hurting financially from both of these and perhaps they are trying to "change the subject" by taking the focus away from hardware and more towards tha ctual user profiles.

Just some thoughts.


@Appdetective - The BS is not just from ex-Softies, but Ex-Citrites too. But then what's the difference?

I don't think analysts care much about the desktop at this stage. VMware's stock has been on a steady climb for the following reasons: Cloud Computing and corporate spending on data center.


I really do not know what to say, so pardon very much for saying nevertheless… As virtualization.info is quoted in the article, let’s quote Paul Maritz in his statement: “…We are providing client-side, big client-side functionality with our offline View capability which comes as part of 4.5. The feedback that we got from our customers is the market’s not ready yet for a bare metal, naked hypervisor, and instead we are supplying essentially a Windows within Windows hypervisor which gives us much better coverage over the installed base in particular”

So, VMWare is thinking that a branch of the very first product will Ace (pun intended) the VDI in some bland XP-mode? I think not. With the Goldman Sachs (also over at virt.nfo) the markets slant towards XenDesktop seems to have tipped over the balance in their favor. Then again, one questions what the market is, or the balance for that sake? How’s it all along the space with real comparative figures vis-à-vis old skool PC – even Client Server? Emerging as in emerging technologies have for me the thought of Emerging Countries…

Yeah, by the way… I’m not a PC guy per se. It’s just that I have a hard time swallowing the VDI cake when fully knowing it’s not whistling to the fanboys melody. Sometimes waking up and smelling the coffee is quite nice, even grand.



"lot of organizations are looking at an end-to-end Citrix solution and are moving away from ESX"

Really ? Have you read the latest VMware financial reports ? They are selling ESX like crazy, and raising license revenues year over year.


Any earnings report will always emphasize the good and never the bad. It simply has to be glossed over. VMware is not making any money with VDI and I suspect Citrix is barely eeking by and this is primarily due to siginificant price erosion and inclusion of XenApp in with XenDesktop.

Back to the topic:

It is about the users. How do you enable them to work from anywhere, gain access to any information necessary to accomplish their jobs, and create a strong technology eco-system?

I do not know.

Throw lots of dollars at the project and hope it works! Yeah, that's smart.

Here's the issue: User want nothing to do with complexity. We certainly do not and we ARE technology experts. Overloaded SME's.

Reduce the complexity and you'll get satisfied users.


VDI is and will remain niche for foreseeable future. The reason being that it requires very specialized skillset that IT desktop teams at many customers do not have. They know how to patch Windows and run sysprep but when it comes to ESX and storage they are lost. Vmware's R&D is better spent on where the money is. VDI market is in its infancy despite all the clueless "industry analysts" who have been predicting meteoric rise of VDI revenues but that has never materialized yet. So Goldman is predicting Citrix will ultimately have bigger market share? Just like they predicted mortgage and housing markets would continue on a roll while selling out to their investment buddies via back door (allegedly, never proven in court but settled with the fed)? Laughable.


Came back home, Quite honest I wanted some ping back on what I raised, then again I thought that anyones plumbing will do allright, Guess it didn't do, guess it wont, Everyones a damn artist :)


I think his statement was intentionally vague. It's interesting that we all seem to interpret it differently.  My opinion is that VMware learned thier lesson that the cost of managing a piece of Hardware is mostly made up of the hardware break fix costs not systems management.  Citrix cleary does not get this.  I attended a Citrix Intel webex the other day and Simon Crosby was actually stupid enough to mention that client side hypervisors are going to reduce the cost of managing desktops which according to Gartner costs $5000 a year.  I asked him how much of the 5000K was hardware and how much was systems management.  Of course he had no clue.  If he or Citrix actually talked o customers he would know that's system management is a tiny fraction of the $5000.    While I'm on the subject  Gartner missed the fact that companies like Dell charge a couple extra bucks to load the corporate image and drop ship a PC for next day delivery which eliminates all those so called break fix costs.

VMware is way ahead of the game when it comes to understanding how useless client side Hypervisors are.  The only mistake they made was not adding any context to the statement.

On the statement that user data is what matters.  They are dead on if you have a good roaming profile solution and app delivery model ( Yeah I know I don't but I'm trying)  then who cares if your in VDI your profile loads and if your on a laptop your profile loads.

As always this is just my opinion. Maybe my interpretation is wrong but that's why we post here!



I am confused on how you differentiate the cost values for hardware break-fix from within all of the systems management costs.

How often do you replace hardware on each workstation? A $500 PC can last a few years with an odd hardware hickup every now and then.

The real problem is Systems Management not relating to hardware.

- Mgmt/Break/fix costs of an OS (patches, upgrades, etc.)

- Mgmt/Break/fix costs of an App (patches, upgrades, etc.)

- Mgmt/Break/fix costs of a profile and mgmt of user data

It is said that hypervisors will make your hardware outlive traditional hardware replacement cycles. This is the same with dedicated SBC because the hardware on the endpoint is virtually useless. It's more of a feature of desktop virt, not hypervisors.

VMware just doesn't make sense to me. They won't even give me a choice to be able to manage my end point, they just want everything in the datacenter.

Citrix is actually working towards more of a choice, if you want your environment to run on the client or on the server.

The client has a huge diverse ecosystem of hardware choices, which makes it harder to manage. Something I think VMware took the easy way out of for now.

It's all about perspective, whether you see use cases for local based computing or only server based computing. IMO, a robust desktop delivery architecture can't survive without the dynamic flexibility of both options.

Also, if your buying a consumer PC from the local store it will come with Windows pre-loaded. I am going to assume that MS will let the vendors load XenClient underneath Windows without hassle, but CVP on the other hand...

The use cases I am currently looking at is strictly SBC with thin clients for our labs. But this is just a tactical solution, mainstream requires local + remote execution via a streamlined central mgmt.


@Icelus - prove it


@watson, What information is Sherlock giving you? It is a well known fact that desktop costs are not CapEx heavy. Ask anybody who has worked for more  than 1 month in the industry and enjoyed patching machines, inventory, software updates etc as @Icelus correctly points out. Ask anybody who signs the the desktop budget how many management agents go on top. Ask them how many people manage the desktop. Come on dude, who doesn't understand that.....

Clearly you have miss the concept of central management and how they apply to Type 1 or 2 client slide hypervisors. Ask anybody who has to manage laptop fleets how hard that is and the expense and hassle. You really believe that VMware understand this........LMAO

Oh and do tell us how much dell charges for 10,000 separate images to drop ship overnight. And then explain how they keep them up to date. Darn that reality thing again.

Also roaming profile is not the data, that's just settings. If it was you also know View has no solution since RTO did not make the cut.


appdetective - enough rhetoric as you always say to everyone else -PROVE IT!


@appdetective - Just so you know my data comes from helpdesk tickets from a large SCCM delpoyment.  it's a couple hundred $ per year at best.  more like $50


@watson. You just did by telling the world you spend money on SCCM and spending hundreds of dollars every year on OpEx when you can buy a really good PC for less than $1000 and depreciate it over 3 years. So your total cost over 3 years is $1000 PC, + $500 per year (software management costs plus people) = $2500. OpEx is > CapEx. Depreciate further and the ratio get's even more. Those are very conservative OpEx numbers assuming your are very well managed, go to many shops that are in either expensive areas or less well managed and costs quickly climb back up to the Gartner type numbers. If you claim $50 a year, I don't believe you. You pay people more than that a year to manage SCCM when you scale out all the other factors. How much money do you spend patching apps, the OS, on logon script changes, GPO policy, printers etc. It all adds up to OpEx to provide a desktop service. And it's not rhetoric, it's fact, your logic is so full of holes that I am simply calling it out. I challenge you to find sane people who believe desktop CapEe < OpEx and I will show you an environment that is completely f'ed up run by people who have no idea what they are doing.....


@appdetective your the one telling me it costs so much so why don't you produce some real world numbers or you know what!



i will make this easy for you.

what management tool do you use?

How maney users do you have?

what's the breakdown of your helpdesk tickets


local software issues?

- corruption

- user error

database software issues?

forgot my password?


I can go on but I doubt you have the answers to this list.  If you do we can debate costs based on objective data as opposed to your marketing BS.


@watson, briefly, hardware costs. Iike I said before, cheap PC about 1K depreciated over 3 years for accounting, reality they stay around for 5 years. Less than 5% fail during that time. More often I have to upgrade them to buy more Horsepower for a new app.  Or I need to spend money getting users to move around to new locations and on DR. Those are huge costs. Software is ESD, patching tools, inventory, config mgmt, security etc. They all cost roughly $150 per machine after heavy discount due to large volume. Then I have to pay maintenance. Then I have to pay people to perform tasks on these machines for maintenance every month, have the help desk deal with ad-hoc requests, troubleshoot app issues 90% due to configuration drift etc. This adds about another $800 per year, per PC on to the cost based on my world which is fairly well managed. In the virtual desktop world I can massively drive up the number of desktop per admin, as opposed to PC guys running around a building. Help desk ticket, tiny % on hardware breakfix which is outsourced anyway to make it cheaper. Software and app issues 90% plus due to configuration issues. As a volume of total tickets that's about 30%. 50% are still around password resets or can't connect and user error and hence trying to build self service there, and the rest are how to questions. Even in portions of the help desk that are outsourced the costs are not really that much cheaper unless there is some level of automation of self service.  Also most helpdesk tickets are poorly categorized and consist of really bad notes, although getting better more recently. So trying to read too deeply into them is often a fruitless exercise. You have to look at staffing numbers overall on the helpdesk and call volume, mean time to fix etc to balance with the numbers. Sounds like you are a helpdesk mgr so I am sure you know what I mean.

That's just helpdesk, I also have to consider the costs of level 2 helpdesk, level 3 senior staff and account of engineering teams who waste time trying to configure the environment. Group Policy, login script change, No way is any idiot helpdesk person going near that. I've probably ignored all the laptop data in this, but that is usually this sucks because this so f'ing slow because patches are coming down, when others on virtual apps or desktops are up quickly. All of this distributed management and architecture leads to tons of overhead and cost. Compared to CapEx it is at least 4X greater IMO. PCs are cheap in comparison. Those are the facts not marketing BS, it's far more complicated than you are trying to assert. Looking at just helpdesk tickets is a one dimensional view into a complex world that is highly questionable data going into in. When you put all the pieces together the costs become scary. Just go try to fire 20% of your desktop staff and see how many things don't get done and watch security and quality go to hell. You talk about SLA's in some of previous posts, but at scale trust me your model falls apart very quickly.

So since you appear to manage your environment at lower OpEx cost than CapEx at 500 users, perhaps you would care to share also? That's not a sarcastic question either. It would be good to share. At 500 users I agree it's much lower OpEx if you are a simple environment. It's a couple of hard coded images. Or very basic app virt. TS is good enough. I've been in SMB environments that are highly complex, and they face many of the same problems. They have one guy to do it all, and the push out stuff, verify it, fix issues etc, is just a model that kills the admin.

It's amazing at that scale that your would risk Atlantis, but good for you. And it also seems that you think you want expensive HA for desktops if a blade goes down as opposed to fair sharing a % of your users across chassis and datacenters etc. Why? If your OpEx is so slow why spend all that extra CapEx in the datacenter.



In my environment from my helpdesk tickets for productivity workers.  I will use your 500 number the total number of users is far north of 50 times that.

BTW I'm not a helpdesk manager but I am paid to make sure our helpdesk is operating at the highest efficiency and lowest cost possible. (I also stay at holiday inns)

Software and OS based tickets for every 500 users 10 per month

Average time to fix - 30 min  total of 5 hours

Systems management software costs including hardware

$22 per user

Application packaging and QA $15 per user

Total cost $41 per user per year.

In response to the high cost of GPO's management engineering etc.  Today these opex costs for VDI are actually higher than for a PC since it's a whole new model.  Add into the mix that even my tier 1 helpdesk team has no idea what VDI is or how it works.  If I don't use persistent Virtual desktops I have to hire another team who knows VDI.

This has not always been the case as it has taken us time to get the right people and right configurations.  But today it's true and I can't change that.  Like everyone else on BM I am a major advocate for VDI and SMB.  The reality is if I'm going to use it for more than a niche use case the costs need to be the same or less than that of a PC.  As far as client side hypervisors go they don't decrease the number of hardware failures because they run on a PC. This is why I don't see a place for client side Hypervisors.

And yes I agree with you it's more complex than I assert but helpdesk tickets and a stop watch don't lie.

To you point of why spend the money on expensive data centers.  I'm not! At least not until I can get the CAPEX cost of VDI down to that of a PC. And get the ongoing environmental opex costs down to something similar as well.

With respect to Atlantis since you seam to work with or at an SMB. Our Atlantis deployment is huge.  I don't so it's very small with respect to the total number of devices.

I'm not going to give any more detail on my operations than this so don't ask.  If you want to prove me wrong give me your numbers in detail.  Again I want to do VDI I just can't justify it at scale.


@Watson - There is no need to get defensive, it is a well known fact that as appdetective states opex > capex.

To prove this point there is no need to post numbers because they are all subjective but lead to the same point, this is an old conversation that has already been done before and studies have already shown the operating costs outweight capital expenses.

The reason why I called you on this is because quoted on VMware's website is the following:

" A typical organization spends approximately 3 dollars in operational expense for every dollar spent on PC hardware acquisition. Reduce your desktop acquisition, operating and management costs through desktop virtualization and improved desktop management."


VMware is obviously trying to spin the fact that they cannot control the end point devices the same way as they control the datacenter, so they are virtually ignoring the devices and focusing on managing the "user" aka VDI instance in the datacenter, then the devices don't matter.

They are ignoring the fact that the end user devices will only get smarter, they should not be ignored. It's a price that any VDI vendor will pay in the future.



I'm not getting defensive - I'm merely stating real world facts and asking you and appdetective to do the same.  Both of you have referenced studies which suggest it costs more to manage a PC than buy one.  I don't know about you but when I make decisions on anything to do with spending money I need to base the decisions on facts!  

Ever hear the statement if so and so told you to jump off a bridge would you?


Everybody was kung-fu fighting....

Watson your users must all be desktop admins, who do their own patches and application upgrades and all have $2,000 laptops.  That's the only way I can think of Cap Ex > Op Ex

Regarding this article....I don't even think VMware understands what they said.  Just like they don't understand the desktop (I said it!).  Just more of the same marketing junk to attract mystery and attention.


nope - Thankfully there not as the cost of managing PC's would be over $10,000 a year!


appdetective already delivered facts, which should more than suffice.

I see no need to divulge any information on the organization I work for to prove a point. Anyways, I have already seen our org and the world agree on the opex > capex discussion.

Let me approach this differently:

I think everyone can agree with your statement that the management costs of hardware is mostly made up of the break/fix costs, actually I believe all the management costs associated to hardware is because of the break/fix costs. I don't disagree with you here. (BTW those break/fix costs are still considered Opex)

That $5000 number, being subjective, is based off of the combination of capex and opex for traditional PCs. The Opex is the management costs of the hardware, software and associated license/maintenance fees as well as IT support. Capex is the initial costs of hardware/software.

I was hung up on your statement at first because I thought you were trying to say that capex > opex. But now that I dig deeper I think you were trying to say that the opex of hardware maintenance is greater than the opex of software maintenance.

The true expenses of IT are actually software driven by maintenance fees and continual IT support. For 1 workstation you pay to maintain there are countless software entities that you have to manage and more can go wrong.

If you want to strip responsibilites of the client away and purchase thin clients to decrease the opex of hardware maintenance for your organization be prepared to increase your capex because it will cost more for you to buy computing/storage resources in the datacenter than to buy computing/storage resources for your end points.

I find this article quite informative:


"When talking local disk vs. SAN for desktop virtualization, the first thing that comes to mind for me is, Why are we even comparing these two technologies? Are we trying to save on CapEx expenditures to justify the project? If so, both Brian and Ron know very well that VDI does not deliver significant CapEx savings. Desktops have no data center footprint and have always been a decentralized model.

Desktop virtualization, however, will save you significant money on OpEx, which is where the highest spend tends to be. CapEx costs have been steadily going down and represent less than 20% desktop total cost of ownership, according to Gartner, IDC and others; meanwhile, OpEx costs have been steadily going up and represent more than 80% of total cost of desktop ownership."


@Icelus. Well stated, the use case as it stands today for VDI is not CapEx, irrespective or your storage design, although I think most of the info week article is BS ignoring the fact that others like @Claudio have pointed out, that SANs configured today vs. tuned for VDI just add to further cost and politics. Those projects will remain stuck and go nowhere for a long time, although I do here Cisco now pushing UCS as a way to reduce CapEx as well for desktop virtualization. As much as I hate lock-in to hardware, may have to look into this at some point. Curious if anybody has a view on UCS for desktop virt.

@Watson. No I'n not going to waste my breathe and give you numbers that are unique to places that I work at, and that you can get from a million different places. I you still don't get the CapEx story, then there is little I can do, except wish you well in you prolonged lab experiment that will stay in the lab. If Altantis is so prevalent in your organization then there is only place you work. (if not you are not large at all). Assuming that to be the case the single large customer that Atlantis quote, it is well known what a cluster F that has been to get right and how much hand holding it has taken to optimize. Sure it may be working by now, but there is no way Atlantis has been able to replicate this anywhere else and hence my comments on the maturity vs. potential of their solution. Besides the customer I am referring to is still not fully deployed, so it's still wait and see.


@appdetective - don't change the subject this has nothing to do with Atlantis.  It's about how to justify moving to VDI/TS.  You can quote any report you want and you can say OPEX is more than CAPEX all you want.  The reality is in a well run IT shop the numbers don't lie.  When I started posting on BM I was hoping to find ways to move VDI/TS  forward in my organization by collaborating with a large group of experts. Now that I have been here for a couple of months I'm finding this is a wasted exercise.



Every post I make gives me an opportunity to intellectually grow. On one hand the validation of opinions of others gives confidence that you are on the right track. On the other hand getting your statements picked apart should show that there are discrepencies with other views.

I tend to learn more if my views are ripped apart and proven wrong. It's just what you are saying is not what my org and the industry is saying.

Keep posting.



I'm all for that however in my humble opinion these discussions should be about facts not marketing hype.  In early posts others have mentioned studies done by VMware and Gartner proving that OPEX is more than CAPEX.  I can't post a link to the recent Gartner report that states OPEX is not as high as they once thought but it's there if you have a login.  

I could have pointed to reports from MS and intel which point out that VDI is far more expensive than a well managed PC.  The point I am trying to make which seems to be lost in the chest pounding is VDI costs more unless the vendors and us customers come up with a workable low cost alternative.  

I welcome the opportunity to have my objective data picked apart by others with objective data of their own which proves I'm wrong.  Keep in mind at the end of the day when I and I'm sure you make a decision we need to choose that which is best for the business this includes the least risk to interrupting productivity at the lowest cost. As to what the industry is saying it's all about what the vendors want to sell today. I can't comment on your org but I can say that I know a good many people who are in similar positions to mine in large organizations who have similar opex costs and have chosen not to look at VDI due to CAPEX



I agree with you that Capex of Desktop Virtualization is larger than Capex of Traditional PC's, because like I said before the computing/storage resources in the datacenter will cost more than the client.

Guess what will bring down the Capex of Desktop virtualization?

1. New datacenter technologies that are VDI aware (ie. storage) to make SBC more appealing for executing os workloads.

2. Allowing the choice of the execution environment, whether it's server based computing or local based computing. Massive benefits can be seen with this choice.

Desktop virtualization makes it a breaze to cut Opex down, however in it's current state the Capex can be more than Traditional PCs. Like I said, I agree with you here but you didn't say that before.

Our context of our posts were only comparing the Capex vs Opex of Traditional PCs.

Bottom line, in the upcoming years client based computing (type-1 client hypervisor) will be coddled out of it's infancy and reigned superior than SBC to be  the major computing environment.

If that doesn't happen then like I have said in my previous posts, Thin clients will become more than niche.

In my opinion, the only way thin clients will be more than niche is if the major desktop virtualization vendors (Citrix or VMware) purchased a thin client vendor and gave them away for free with their user licenses.

Oracle has it right that they are using their sun ray thin clients with their VDI solution (but charging for it like crazy), however they are missing every other piece of the desktop virt puzzle.