An alternative to Terminal Server: A case study for hundred of apps and thousands of users.

Gabe and I were working with a customer-a university-and we ended up recommending a VDI solution instead of a Terminal Server-based solution.

A very strange thing happened yesterday. Gabe and I were working with a customer—a university—and we ended up recommending a VDI solution instead of a Terminal Server-based solution. Afterwards I was feeling, “Wow! I can’t believe I just did that!” But I really feel it makes sense. And in fact I think it might continue to make sense more and more, and now I’m wondering if VDI can start to come out of the niche and into the mainstream?

Let’s start at the very beginning. Gabe and I worked with this university six months ago. They were not using any server-based computing or streaming or anything like that. It was a brand new environment. They had four scenarios (or “use cases”) they wanted to enable:

  1. There are 1200 lab workstations throughout campus. Users need to be able to walk up to any one of them and access any of 200 applications. The users also need access to their own data and profiles.
  2. They want to publish a remote desktop via server-based computing to people so that they can access the “lab workstation” from their dorm rooms or off campus.
  3. They want to publish individual applications (as opposed to a full desktop like in Scenario 2) to users on their own computers.
  4. Longer term, they want people to be able to run these applications locally on non-university-controlled workstations (i.e. student laptops), and they want this to work offline. 

The initial plan

For Gabe and me, these four scenarios were perfect for a combination of traditional Terminal Server-based application delivery and application streaming.

We were thinking they could use something like SoftGrid to isolate and stream all (or most) of their applications. Then they could add some Terminal Server and a third-party application publishing tool to deliver individual applications. Our initial suggestions for each scenario above were:

  1. Use SoftGrid to stream the applications so they run locally on each lab workstation. Install the few non-SoftGrid-compatible applications natively on the workstations.
  2. Use Terminal Server, along with Citrix Presentation Server or one of the cheaper alternatives, to publish server-based computing desktops. A combination of SoftGrid and local installs would be used to get the applications onto these Terminal Servers, much like the lab workstations.
  3. The same Terminal Servers, running Citrix or whatever, can be used to deliver seamlessly published server-based applications to desktops and laptops.
  4. For the applications that can be sequenced with SoftGrid, they could also be streamed to Windows clients for local offline execution.

That was our recommendation and plan six months ago. Let’s look at how that worked out.

Six months later: the reality

Looking over our plan, you can see that most of it hinges on SoftGrid, and getting as many applications sequenced as possible. After six months, the university was only able to sequence about 100 of their 200 applications. What prevented them from sequencing all of their apps? Several things, including:

  • Problems with interoperability. Lots of applications require IE or Office components, and those can’t work outside of their bubble.
  • Problems with interdependencies with the local OS. Many applications required Java, the .NET Framework, or IE plug-ins to be installed locally on the host machine.
  • If you virtualize IE to get it to work with the plug-ins, there are problems presenting that to the users since you can’t really remove their own local IE.
  • Browser plug-ins are not designed to be managed centrally, so they’re not really updateable centrally. But if they’re virtualized the plug-ins, then that means that they’re managing them, and they can’t figure out a good way to do that.

At this point, the university has sort of given up on SoftGrid. They are happy with it for the 100 applications that they’ve successfully sequenced, but they don’t feel that it can be the solution for everything. And since SoftGrid was kind of the keystone of this whole plan, they called us to discuss the strategy.

We talked about a lot of stuff. We talked about some of the tactical issues around SoftGrid sequencing and some ways that they can address some of their challenges. But we also talked about everything that they could do from an architectural standpoint. We discussed VDI, Ardence, VMware ACECitrix Presentation Server, Citrix Desktop Server, Provision Networks Virtual Access Suite, the upcoming VMware OnDemand streaming, and hardware pricing.

Fundamentally, the university had two requirements:

  1. They want a solution that can support the four usage scenarios outlined above.
  2. They want the solution to be easier to manage than the manual way they do everything today.

As we talked through the various options, the university really latched on to the idea of VDI. Specifically, we talked about how Provision Networks Virtual Access Suite has a feature where you can seamlessly publish applications from individual VDI VMs. We talked about how Ardence could let all of their workstations—physical lab PCs, VDI VMs, and local VMs—boot from the same disk image.

An overview of the new plan

When we were done, the university had a very different solution set to support their four use cases:

  1. Lab workstations: Boot from a single Ardence image. That image will contain as many apps as possible via SoftGrid, with the rest as local installs.
  2. Published desktops: Use a VDI solution based on VMware VI3, using Provision Networks Virtual Access Suite as the connection broker and web interface. The VMs will boot from the same disk image as the lab workstations via Ardence.
  3. Published applications. Use the same VDI back-end as in Scenario 2, also including Provision Networks and Ardence. Provision will allow the university to publish individual applications via the web interface in a seamless server-based computing way, except the users will connect 1-to-1 to a VM running Windows XP instead of a Terminal Server.
  4. Offline use: Use VMware ACE, also booting from the same image that they use to create their Ardence image.

Details of the new plan

Now that we’ve highlighted the new plan, let’s take a more in-depth look at the elements that make up each usage scenario.

Lab Workstations: Ardence

This was the biggest “no brainer” part of the design. They have 1200 workstations around campus. They have a fast network connection, and all of these PCs are more-or-less the same. Right now they’re managing them using traditional imaging techniques. Moving to Ardence makes total sense since they can have one single instant image for all 1200 workstations. That Ardence image will contain as many apps via SoftGrid as it can (all pre-cached), and the rest will be locally installed.

Published Desktops: VDI and Provision Networks

Here’s where it gets weird for me. Traditionally I would be in total support of using Terminal Server here (most likely with Citrix Presentation Server). I’ve been convinced for years that VDI just doesn’t make sense for general SBC environments. My two main problems with VDI are:

  • VDI means that you have to manage a separate disk image for each user. This means patching and fixing and all sorts of headaches.
  • You can fit many more sessions on a Terminal Server than you can Windows XP VMs on a VMware server.

However in this case, I don’t think these two problems existed.

With regards to managing disk images, Ardence completely solves that. And the university intends to use the exact same Ardence image for their VDI environments as they’re using for their lab PCs, so really there is no additional management.

Regarding user density on servers, I did some research and have some interesting thoughts. For years, a “standard” USD$3,000 Terminal Server has had dual processors, 4GB RAM, and two 15k mirrored drives. Even as processor speeds have increased, the number of users has remained about the same. (Of course this varies widely, but it’s generally in the 50-75 sessions per server range.)

That same server, running VMware with Windows XP VMs, could probably run about 15 VMs. This means that you would need 3 or 4 times as many servers to support the same number of users, when comparing VDI to Terminal Server.

In the past few years though, server hardware has really changed with the introduction of multi-core processors. In fact, today’s USD$3,000 server has four cores (two dual-core), 4GB RAM, and mirrored 15k drives. But what’s interesting is that server still only supports 50-75 Terminal Server sessions, even though it has twice the processing power.

The reality is that 32-bit Terminal Server just doesn’t scale that well. This is not the case with x64 Terminal Server, but that requires a lot more memory to run the same 32-bit applications.

If yesterday’s dual-core $3,000 server can run 15 VMware VDI Windows XP instances, then today’s quad-core 8GB RAM can run 30. And for $6,000, you can buy an eight-core (two quad-cores), 16GB RAM, four 2.5” 15k SAS drives server which can run 50 simultaneous Windows XM VMs.

Even though this server is twice as expensive as what you’d use to support the same number of Terminal Server sessions, there’s one huge benefit: If you use it with Ardence, then it’s running the exact same disk images as your workstations. You don’t have to worry about Terminal Server application compatibility. You don’t have to worry about x64 compatibility. You just hook up your users and go.

Published Applications: Provision Networks from VDI Windows XP Virtual Machines

Seamless published applications is Citrix’s traditional strong point. And in a scenario like this, I have traditionally always recommended Terminal Server-based solutions extended with something like Citrix Presentation Server.

But in the case of this university, if they already have one single disk image for their lab PCs and their VDI published desktops; why not use this for published applications too?

Even though you’ll need a $6000 server to support 50 users instead of a $3000 server, the fact that you don’t have to built out a whole separate architecture and test all your applications on Terminal Server will save a ton of time and probably create a quick ROI on those more expensive servers.

We recommended Provision Networks in this case because this is not something Citrix can do. Presentation Server can only publish seamless applications from Terminal Servers. But Provision Networks can hook into their VDI broker and connect users via their web interface to seamless published applications, even if those applications are coming from backend Windows XP workstations running on VMware instead of applications running on a Terminal Server.

Offline Applications

The final aspect of this solution is the offline use case. As I’ve written previously (here and here), Ardence does not work offline yet. However, VMware has (and has had for years) an offline solution called “ACE.” ACE is basically a VMware hypervisor, a disk image, and all the configuration details needed to run it packaged together into a centrally-managed environment. ACE even handles checking in and out licenses to ensure application compliance in non-connected use cases.

Due to the network requirements, the university figures that users will have to be onsite to receive the disk image to take it offline later, although they’re also thinking about distributing the images via DVD (or even eventually with VMware’s OnDemand streaming technology.)

Final Thoughts

Like I said, this whole concept seems strange to me. It seems strange to recommend a solution with no Presentation Server and no Terminal Server. But I think this is the right thing for this university. The proposed solution lets them manage a single disk image for their entire environment. They can use flex profiles and policies to control the user environment.

Why no Citrix? (Well, other than Ardence, which is now owned by Citrix.) The problem with Citrix in the server-based computing market is that their desktop server product is a completely separate product from Presentation Server. Even when Desktop Server version 2 comes out, it’s still a separate farm, a separate database, and additional licensing on top of the $500 per user or whatever Presentation Server costs these days.

Provision comes in with their single product at something like $100 per user which supports Terminal Server-based and VDI-based SBC models in the same product, and it also provides the seamless application publishing from Windows XP VMs which is perfect in this case and not even on Citrix’s roadmap. So Provision is a no-brainer.

As for using VDI instead of Terminal Server for the published desktops and applications, this means the university gets broad application compatibility and can use the same desktop images everywhere, and it only costs them USD $6,000 per 50 concurrent users instead of $3,000. Really that’s not too bad from a capital cost standpoint when compared to the fact that they have a much easier time managing the thing and they don’t have to figure out local solutions and more servers for non TS-compatible apps.

By the way, someone pointed out the fact that this VDI solution would require a $3,750 VMware VI3 Standard Edition license for each $6000 server. This is true. Then again, you don’t need a Windows Server license for each of these since you’re not running Windows Server. That Windows Server license would be $1000 if you have up to 4GB of RAM, but $4000 for Windows Server Enterprise Edition if you go above that. (Which is common and in fact more expensive than VMware anyway.)

As a final thought, when discussing this whole thing with Gabe last night, he pointed out this is most likely the exact reason Citrix bought XenSource. Maybe Citrix realized VDI could be a bigger threat to Terminal Server-based SBC sooner than we did? Maybe Citrix realized that hardware will scale up better with individual Windows workstation VMs, and that applications will be more compatible and more portable, and that Ardence will take care of the disk image problems.

If so, then bravo Citrix! Can we just please have Presentation Server and Desktop server integrated into a single farm in a single product? Can you integrate it with Ardence and XenSource? And can you make it cost less than $1200 per CCU? If not, well, whatever. This SoftGrid/Provision/VMware/Ardence solution looks pretty cool.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Totally agree Brian, Citrix and Presentation Server really hasn't moved on significantly for years - sure printing is improved and you can mange settings a lot better but they don't even have a single management interface yet! I mean how hard can that be to get right.

They really need to get the VDI, OS imaging and PS interface tightly integrated into 1 seamless product and like you said the price really needs to start dropping. When Win 2008 comes out and VM get there VDI solution up and running. Citrix will really start to loose customers. 

I mean are most of Citrix customer interested in NetScaler and EdgeSite? Not really is my guessing certainly not for there price



Just thanks for sharing your thoughts in this customer case scenario. I can feel this is a nice session for BriForum Chicago / Australia next year!. It's interesting to see that a combination of more and more solutions is really powerful. What a great marketspace!



For just there VDI solution, Provision charges just $50 per concurrent user.

You failed to explain why Terminal Server/Citrix didn't work.   You've only explained that Softgrid failed, not TS/Citrix.  If Softgrid failed, then you will have the same challenges with VDI as a Terminal Server.
You won't need the costs of Presentation Server wlicense hen Desktop Server 2.0 is out.  So forget the $500 (actually $450) as it will have dramatically lower per user cost. 
Citrix Desktop Server 2.0 cannot serve individual published applications (Scenario 3), so it won't work.

The problem as you know is that not every application is TS compatible. And for these 100 "other" apps, if they're tricky enough that we can't get them to work in SoftGrid, then there's a good chance they also won't work on TS / CPS. So really, what are we gaining by using TS/CPS? It would be a whole additional architecture to support, not to mention $450 per user in addition to what we have.

Instead the proposed scenario just leverages the existing technology investments. I mean sure the server to support 50 users is $6k instead of $3k, but the savings in simplicity and not having to make apps work in TS is well worth it.

  • Problems with interoperability. Lots of applications require IE or Office components, and those can’t work outside of their bubble.
  • Problems with interdependencies with the local OS. Many applications required Java, the .NET Framework, or IE plug-ins to be installed locally on the host machine.
  • If you virtualize IE to get it to work with the plug-ins, there are problems presenting that to the users since you can’t really remove their own local IE.
  • Browser plug-ins are not designed to be managed centrally, so they’re not really updateable centrally. But if they’re virtualized the plug-ins, then that means that they’re managing them, and they can’t figure out a good way to do that.

This is the list you provided for issue with Softgrid.  Each of these items are addressable.  Applications CAN reach outside the bubble not into one.  Office and IE should be installed locally anyways.

Java, .NET and IE-Plugin CAN be included in the sequence.  It NO longer required to have these already on the local machine.

The third and fourth complaint do not make sense...  Yes, you cannot sequence IE, but you can run IE inside a bubble.  You can install the IE plugins and run then inside the bubble and that works great.

It sounds like your customer does not really understand (technically) Softgrid and how to make things work.


It would be great it you could post a "post mortem" in 6-12 months from now to see how the implementation of this went. As you said "This SoftGrid/Provision/VMware/Ardence solution looks pretty cool" but I think we'd all be interested to see how it goes in real life...especially with the key technologies being provided by no less than four different vendors.  This doesn't sound like something that "just anyone" would want undertake without real good reasons (which in this case there do seem to be).


50 Windows XP VM per server... I'd like to see that... You'll be lucky to see 1/2 that.

If you really want a killer solution.. You shoud use Virtuozzo with a Datacenter license.  It would be MUCH cheaper and you'd be able to scale up much better than VMware.  You could get 100-200 VMs per box compared to a piddly 25-50 with VMware. 

why not, it seems pretty straight forward.  For years, we cobbled up TS, CPS, SoftGrid, RTO, Aurema, some UP vendor, some profile management vendor, RES or Scriptlogic, and half a dozen other vendors, but I never heard anyone complaining....

SBC isn't a panacea.

Software Virtualization (i.e. SoftGrid) isn't a panacea.

VDI isn't a panacea.

Ardence isn't a panacea. 

Anyone who fools themselves into thinking that a tool like SoftGrid or any other tool is going to be the single solution required for all of their applications is fooling themselves.  They're also fooling themselves if they believe SBC will fit the bill.  They're also fooling themselves if they believe VDI will fit the bill.  The reality is that there's a wide range of applications out there and most of them are crappy.  Thank god for that though or I wouldn't have a job. ;)

I've seen customers go down the VDI path because they didn't want to deal with the complexity of infrastructure buildout and we're happy with success criteria of getting their apps on SBC.  Many of those customers that went the VDI route believed that they'd solve their issues by sticking with the same ways that they deliver software to their desktops, but do it in a VDI mode.  The reality is that there's always applications that fall outside the realm of something that's practical in a remoted environment.  There's device interaction issues, there's complex UI/graphical elements, there's bandwidth issues.  It doesn't solve every problem, which brings me back to my favorite quote:

"When all you have is a hammer, everything begins to look like a nail"

Customers that believed XXX technology was going to be the approach they'd use to solve world hunger quickly realized that life just isn't that black and white.  Furthermore, while the multi-vendor approach may yield some additional capabilities, it also introduces a higher level of complexity and increases the difficulty in troubleshooting your application delivery scenarios.  Next, as already mentioned the SoftGrid failures that you mentioned are known quantity and are things that can be worked around (to some degree - the IE one is messy unless you have ways of forcing specific virtualized IE versions in all cases).  Finally, the mixture of proposed vendors that seems at face value to fit the needs of the client will result in a fairly complicated set of usage criteria when it comes to online well-connected, online poorly connected, and disconnected modes of operation.  Disconnected users is one area in particular that makes this architecture particularly challenging.  Though I suppose it really depends on how much the IT department is concerned about user acceptance.  I've been to many environments that you can't rely on the user to know which apps can be launched in an online well-connected vs online-poorly connected vs offline state.  Solve that issue either through complex IT solutions or through thorough user training and you've got that issue nailed.  However, I haven't seen anyone do that well....yet....




Citrix presentation server works fine in publishing apps those are supported on windows 2003 server, we have number of apps that are not supported on windows 2003 like, Auto Cad, Adobe CS3 suite, JDA etc. Even I talked to these vendors they are not interested or plans to support on Win2k3, that mean, we can't publish these apps via Citrix. Citrix PS server is not a solution for all apps, so we have find other solutions.

On the other Softgrid also has its limitations like sequencing Adobe CS3 is not possible so far, Office2007 have some problems as well. In that case VDI could make sense.




Thanks for sharing these thoughts. I've seen similar setups at other clients. Thankfully, the university has worked with images before. I say thankfully because imaging, like Terminal Server, requires a different mindset. If you are working with a single image that everyone uses in multiple ways, you'll have a hell of a time regression testing for every single change. Also, if you plan to sequence 100 applications and locally install say the other 100 you could very well be going into the application compatibility nightmare as you would on Terminal Server.

I second that request for a "post-mortem", it's always great to hear how the plan became reality.

Michel Roth


I guess this is the new world after all .. :) I think we have all seen TS/Citrix combined with other vendors. The day had to come for the mobility world to get completely different option :-) I am sure as Shawn Bass pointed out, there will be other challenges with this setup, but i am sure it would have been the same with a more traditional approach.. Very exciting project, and a "Post-mortem" would be excellent :)

I am sure all that you are asking for from Citrix is just going to be included in the normal SA for existing customers and part of the Platinum License for new customers ;-)



Selfproclaimed expert @ UpcomingGrasshoppers.OnTheLose.... .com ;)

Brian, nice solution, and indeed I am interested in the post-mortem. I am wondering about user experience in this solution, regarding graphic intensive applications that will become more and more prominent (think WCF). How do these applications look and feel by using a VMWare VDI? Will they perform to user satisfaction??
You know? I completely forgot about Virtuozzo.. But great idea! Do you know where they fall price-wise? I'll also have to see if Provision's VDI broker can hook into them.

Remember that VDI is still server-based computing, it's just 1-to-1 instead of 1-to-many. So of course, WPF and graphical applications will perform no better or no worse in VDI environments than TS environments. In the future, Microsoft may choose to "remote" Windows at a different layer, perhaps below the GDI / graphics interface layer today, and maybe more towards the window manager or below WPF.

Regardless, if this is a problem, that's where local app execution shines. (Local Ardence, VMware ACE, etc.)


First of all. Brian, great topic. THis is really something to talk about. Secondly, I full agree with Shawn's reply. Third. In this case of universities spending money on and OS sounds strange. Let me explain: since years I've been visiting companies and they all said that if there was an option to go to Linux or whatever to save money they would. Eventually even managing a Linux environment can cost more then .... But OK. I truely believe one single solution is not enough to administer and manage an environment. But I do believe that we should not think of distributing a desktop... ;-) Back-end is fine, but front-end, ahum....

So, Brian, dude, think of that as well... ;-) Forget the desktop, go for a "Workspace".

Btw. great article.


That they are the guinea pig for this solution.  Lets cobble together 5 different vendors solutions because it seems cool to do so.  Who cares that some of these vendors are small and could close tomorrow and some of these vendors are huge and cumbersome to work with.  I am sure the university will love working through the support options with each of these vendors!!

This is what the game plan sounds like:

Ex:  Did you try Virtuozzo?  No, I didn't!  Sounds like it might work. . .I'll try it!  Lets throw it into the mix.

Are you saying that Citrix is going out business after foolishly spending $500M (almost half their cash) to try to compete with Vmware? 

Agree 100%.  We have been using SoftGrid for 4 years and I have seen everyone of these cases and each one is somewhat easily addressable and fixable.

Sounds like to me they need a SoftGrid "Guru" in there to give them some "from the trenches" sequencing assistance vs. classroom sequencing stratigies.

I have sequenced 100's of applications and only 1 has ever failed (cept the COM+ / DCOM apps). 

It's technique...



Two questions:

If small vendors "could close tomorrow," and huge vendors "are cumbersome to work with," then what vendors should we be using? Is there some kind of vendor size sweet spot? What is that spot? Is is based on revenue? Employees? Customers?

Second, so if I've never tried something, I should stay away from it? How does anyone expand their horizons then? 


Oh, and one more thing:

You write "Let's cobble together 5 different vendor's solutions because it seems cool"

Are you suggesting that this solution can be done with less than five vendors? If so, I'd love to hear it! But look at the 4 Scenarios the university has outlined. Can we use fewer vendors and still meet all these requirements?

In fact, I would argue that all IT environments are "cobbled together" from multiple vendors. Microsoft for the OS, Citrix for app delivery, someone else for monitoring, someone else for backup, someone else for management... I don't think the "one vendor" solution will ever exist. So we use multiple vendors because we have to build a technology solution to support certain business requirments, NOT because it's "cool."


Virtuozzo lists at $2500 for 2 processors, and their management tool runs about $1500. I think one of the reasons why they are so competitive is that they are very big in the server hosting space, which is quite price sensitive. I would also think their product would have to pretty solid to survive in that market. I saw their new AJAX based management client at VMWorld and that looked very nice. We have one server running Virtuozzo now and are seriously considering it as the production solution for virtual desktops. There are a lot of advantages if you are happy with having your users run Windows Server instead of XP/Vista. I think of it as a middle ground between TS and a full VM. 

Provision can also serve Virtuozzo desktops nicely. The Provision demo has some Virtuozzo desktops if you want to check that out. From a cost perspective, Virtuozzo has an advantage over ESX because you get much higher density per box and you don't need anywhere near the same amount of storage for the virtuals. The smaller VM's are also easier to work with.

We're also using ESX, but this might end up being only for servers. I suppose we could reconsider on the basis of having fewer vendors and the associated moving parts if their new VDM offering is competive. In any case, I think it is worth your time to talk to Virtuozzo.

PS - Virtuozzo even has an application templating feature that might be of interest.
I hate to keep beating the same drum..... This makes me feel so un-cool. If the solution can be simple and elegant doesn’t that outweigh everything else?I think everyone is bothered because it’s rare for VDI solution to come alive that isn’t a Frankenstein with a lot of vendors involved. Everyone hates those guys right? J who wants to deal with more then a few.University or college environment is a perfect place to try new things but practicality has its merits that pay off in the long run. To answer Brian’s question, I think the vendor should match the customer… so make the solution completely open source in this case J now that would be a good guinea pig.



here is a link to the Virtual Access Suite (VAS) for VDI 

At VMWorld we were told that the pricing is $50 per concurrent user. If you want to do TS as well, you double that. 


Lab workstations: Boot from a single Ardence image. That image will contain as many apps as possible via SoftGrid, with the rest as local installs.

Maybe I missed something here, but it seems that you still  have 100 apps that would not sequence and need to be locally installed? I still see pain on the horizon trying to get all of these apps locally installed and working nicely together on the same Ardence disk image. I'd guess that you might end up with 10/15 separate disk images which support each application set that plays well together (starting to sound like silo's to me :-)) and now how do users get at each set of apps from each different disk image if they can only boot one image or take one image ofline using ACE?

 And what about maintaining updates to these 100 locally installed apps?  I can see a new disk image (for each of my approximated 15 desktop images) being cut every week to keep on top of new relaeses and patches etc.

What you have sounds like a great use of available technology, but I'd put some $$$ down that Terminal Services and maybe Presentation Server, and specifically published applications, will play a part in this deployment by the time the dust settles, if only to provide access to some apps which need to be made available across disk images.



In my working with customers, we have found that most shops can get 5-10 VDI VM's/core.  The economies of scale continue to get better as core counts increase and as Intel/AMD make the chips more optimized for virtualization.  Concevably, on a 4 socket 4 core box, you could get up around 112 VDI sessions if go with a middle-of-the road 7 VDI sessions/core.

Also, when a single user's app starts chewing down resources it would typically slow down everyone on the Terminal Server/Citrix server...once your on that physical server your stuck with all the other users on that TS/Citrix server.  In VDI, when a user's app start to chew down resources, the load balancing in VMware evens out the load and makes room for that VM without affecting other users sessions via DRS/VMOTION.  That said, we typically configure the shares (built in) to ensure that a single user's VM cannot overpower everyone else and hog the all the resources on the server.


Just a few thoughts and experiances from the trenches

I'm curious about the challenges they had with their application set. Could it be a lack of packaging / SoftGrid expertise? The reason I'm curious is the worst success rate I saw with SoftGrid was 3.1 and we were still getting 90% success rate with an energy company that did everything from engineering plants, refining, drilling, to retail. I can say on the 4.1 code I'm seeing all the implementations go above 95% success rate. I'll admit SoftGrid isn't suited for every application situation but having about a 50% success rate does concern me and the issues listed don't sound like issues I see. I'll admit strategy is key to sequencing successfully but I don't see why Office, IE and .Net are issues.

On the spot Kevin, successes are booked with smart thinking when (re)packaging. Tools (Softgrid, Citrix, etc.) to deploy are second thoughts. Switching from solution to solution is only increasing costs.


They already have the 1200 lab PCs with 100 SoftGrid apps and 100 local apps. So yeah, it's a pain to make that image work, and yeah, they're going to have to update it a lot, but the idea with Ardence is that at least they only have to update it once.

So that single image is what enables this whole scenario, ACE or VDI or whatever.

But my question is about TS / CPS.. What does that buy? Becuase we'd still need a local element for Scenario 4? Although it would be cool to build a TS and serve problem apps via SBC for Scenarios 1, 2, and 3, but of course if we did that we'd just use Provision since it can create a single farm to manage the apps published on Terminal Servers and VDI images. 


iForum's a couple of weeks away, and the XenSource deal will be completed, MS have just released their preview of Viridian on Server 2008 RC0.

 I think we're going to hear some interesting announcements at iForum with a real focus on desktop virtualisation.

It's about to get as interesting as the early days of WinFrame



You stated that is the end all, be all solution for this university without even testing it!!  And then suggested adding Virtuozzo, well, just because!  This is why anyone should feel bad for the university.  Also, if something goes wrong with this configuration, who are they going to call:  YOU?  Citrix (Ardence)?  Provision Networks?  Microsoft?  VMware?  Virtuozzo?  The hardware manufacturer of the servers?

That is the point in the eariler post about support.  Yes, people go with heterogenous implementations all the time, however, usually they have a standard (or try to) for some portion of it.  Your solution would be very hard to troubleshoot and manage effectivley because there are so many moving parts.

I hope it works.  As you said it, ". . .looks cool" but I wouldn't want to invest my money and reputation in a cool solution but one that worked well, could be managed effectivley and supported without a hitch.



Ardence seems to be the way to go for workstations but why don't you consider using Thinstall to package the apps to a "shared" app folder - avoiding any local install problems and/or distribution issues?

As I see it Thinstall would be the perfect combination together with Ardence: One OS image for all computers and one single .exe file per app for the entire environment...

 Just a thought...

 Tord Bergset

I have been looking at Thinstall and I believe it does solve your problems and as a Platform it can be used for multiple issues beyond streaming apps.
This is why Provision Networks partnered with HP and IBM.  One of the two global giants can put the solution together and support the whole kit.
Brian - great article!  One question though...did you have any issue flipping your Ardence VDisk to an ACE Image?  I've tried this myself trying to get a Ardence VDisk to a  dedicated VM (i.e., not streamed) using converter, and LeoStream.  Everytime the VM reboots it loses it's domain membership.
You can actually get 60 XP VM's with 32GB of RAM with a little room to spare....
I still don't understand what the alternatives are? If you don't agree with this solution, please outline what you think would work here and still meet the 4 usage scenarios.
Well sure, there are alternatives to SoftGrid out there, such as Thinstall, Altiris SVS, Endeavours, etc. Not to start a religious war here, but I'm not sure whether these would be any better or worse then SoftGrid though.

Provision Networks is just one piece of this jumbled solution.  That is the point.  HP and/or IBM could not support the whole thing.


Read again what was written.  Don't be so deep in the weeds to try and pretend you don't understand support.  You used to be a pretty good consultant at HP.  Remember your roots.

True TS has compatibility issues, but the examples you quote are not experienced as standard by TS. Even if there are multiple infrastructure component version dependencies that could cause clashes then a lighter weight isolation like AIE can be used for those specifics (OK I know manual setup on AIE may take time if you've not got a grab bag of rules - siloing then :). Some CPS in the mix would probably add to overall efficiency.
Wow, this is going to be a nightmare to support. Especially for a university that cannot afford a top flight support staff. At least the cost of downtime is low compared to a real enterprise customer.

Ohhhh... Provision Networks got some BusDev person inside IBM and HP to do a press release that says they "partner".... That automatically means HP and IBM will compeltely support this solution. try calling IBM Tech support and ask them to support a provision networks problem. I will bet ya the only response you get is "Provision who?"


Do you really think IBM and HP will provide tech support for every software company that signs up for a partnership? have you ever been to one of their partnering events? They both have hundreds of ISV partners. Only the strategic alliances come close to getting tech support from HP and IBM. Do you really think Provision networks qualifies as "strategic"? 


Wow, I didn't realize that merger and Aquisistions expert from Wall Stree posted here as well. Please, tell us your thoughts on how this purchase in the hottest field in IT is bad deal for Citrix, and what effect it will have on current revenue streams, cash on hand, and future earnings per share.



I love the thought idea that you experiement on paying customers... That is awesome. You have got huge brass ones man!!! Milk em for every penny!
Brian great post above as its obviously sparked some serious discussions/thoughts.  I agree mostly with Shawn B and Kevin K. I think the 100 applications aren’t that bad for first crack, with some expert assistance I’m more than confident that number will go up drastically. From there things that don't sequence properly could be standardized on and the other forwards/backwards versions can be sequenced within their respective "Virtual Environments - or what everyone refers to as sandboxes". We all know SoftGrid isn't perfect but with the roadmap in place and Microsoft behind the wheel, those final missing pieces are just around the corner and it’s just a matter of time for the realization of true SaaS to be achieved.It's so funny I was on a call with a customer today discussing SoftGrid and the customer is looking to provide access to applications on a locked down OS to external partners, and found myself thinking of the scenario you had described above. I even went as far as recommending they take a look at VDI as their solution coupled with a few of the technologies above.As is the case with everyone else, please keep us updated on the post-mortem implementation. Also let me know if the customer would like to have a deep dive on SoftGrid and maybe plan out some implementation strategies. Higher Ed has always been a sweet spot for application virtualization especially with it's off the shelf applications.



Just to be clear, this is a "proposed solution," as mentioned in the article.  True, it hasn't been tested, but proposed solutions are not.  There was never any recommendation to jump right in and do all of this, but given the current state and the desired state that the university outlined to us, this solution could fit all their needs in a practical, econmical, and manageable way compared to what they are currently doing.  Now it's time for a proof of concept to see if it will work or not.

You speak of "so many moving parts," which is true, but compared to the current number of moving parts, it's not nearly as bad.  There may be a way to shoehorn one broad solution into this environment so that it can be managed from one place, but then you risk taking away practicality and economy.

We don't sell any products.  We're not vendors or distributors.  Our job is to find the solution that we feel is the best for each particular use-case, just like any other consultant.

In all sincerity, I would love to hear another solution.  You know what we know - it's all in the article.   If anyone wants to try, shoot me an email - I'd be glad to lend a hand.

We actually talked about AIE, too, for that very reason.  In this situation, though, the university doesn't currently have any Citrix products.  In order to get it, they'd need to bring it in, at a pretty substantial expense.  At that point, there's also no guarantee that the apps they've already sequenced (or even the remaining apps) will work with AIE, so we're also probably talking about keeping SoftGrid around.  There's an awkward trade-off there.
Well, I guess we can agree to disagree here and leave it at that. I'm not really sure how this is an experiment, but whatever. And milking? We don't sell these products, and we don't have any relationships with any vendors. We're just looking for the simplest solution that can still meet their business requirements. And what I still haven't heard from you, is what solution you would propose that can still meet all four usage scenarios? If you don't like this, then don't criticize it... tell us what you'd do instead!

Did you change the ADS Group Policy about the computer password?, Ardence needs to control that functionality.

Looks like an exercise from an interview -- looking at the thought patterns with no real plans for implementation. Is that University looking for full time employees?
Don't do it - that place is a mess.
Dr. Sasha
Why is this place a mess? And what's wrong with no real plans for implementation? At this point we've very early on.. Sort of a "here's our situation.. let's brainstorm and think of all the possible solutions." That's why everything's on the table at this point, and that's why we're not afraid to suggest new ideas.

Thanks for the article. As a consultant, the model you recommended is the same converstaion we are having with customers all over the country. There seems to be a lot of people that say this is too complicated. My response to that, as a person that has sat listening to 1000s of companies tell me their current model, is that is what is to complicated. They can't react to the business needs today becasue of the current tradional way desktop and laptops are rolled out. They have huge IT staffs running around fixing fires and no way to break the cycle. Anyone that does not belive that the model they put forth in this article, is not in tune with the changing business requirements customers are giving us today, is not in synch with the changing IT enviroment. All of the technologies recommended are proven companies, IE: Softgrid, VMWare, Ardence, Provision, etc. This new model is going to allow organizations to deliver what their users need, when they need it and whereever they need it.

IT skill sets are going to have to change to run these models or they will find themselves selling at starbucks.

Granted, there are too many vendors here and this will be a solution today for those companies that can cross the chasm. In the future, companies like CItrix and VMWare will have this solution tied up as one solution. 1 - 2 years from now.

Welcome to the future. Did you think it wasn't gong to change?
In the world of posting comments to a blog entry, such as we have here, it amazes me how many people use them as their own private soapbox to sound off about what they think they know or what they read, or at least their interpretation of what they read. Given that most of the comments are somewhat productive in nature, I am focusing on the other 10%.

Brian and Gabe were presenting a case scenario and provided their innovative thoughts on what they feel may have been the best solution for a client. I do believe that Brian and Gabe did ask for input on what the readers may think is a good solution for each of the use cases, but I know they did not ask for unproductive criticism or off-topic tangents. They are consultants, just like alot of us, and believe it or not, we all have out own biases and opinions on products, support and even clients.

Although, I would say that as far as mainstream consultants go, Brain and his team are in the top tier of all that I have known or worked with throughout my career. So as far as the comments implying that they are not providing thier best solutions for their client, I ask this - do you even know their client as well as they do? How many clients have you suggested solutions for only to find out that, well, uhm, maybe that wasn't the best one? If you say none, and you have been working in this SBC space for awhile, your not telling the truth.

One last thing I need to ask of Brian... can we change the policy of these comments to disallow the "Guest" postings... hiding behind the anonymity of that user does not give some of the commenters due justice.

Great piece, Brian... waiting to hear how it shakes out.
Don't worry Brian, half these posters have never heard the term "Development". Or even know what a Project life cycle is... Otherwise they wouldn't be mentioning implementation. If anything, you're on the right track as you've already done the Analysis/Requirements gathering and an initial design.. The next step would be development/validation of the design NOT implementation. The doctor needs to go back to school.
Thanks for the write-up Brian, definitely makes me think about 'options'.
I was scratching my head a little bit in seeing how a VDI could better handle application conflicts then TS, but your comment about using Ardence and only having to work out those conflicts on one base operating system - instead of potentially hundreds, makes sense.
I would be surprised if they couldn't sequence more than 95% of their applications using both SoftGrid and Thinstall. (yikes, another vendor ;). Any solution that's derived for these objectives will require multiple vendors. Heck, I can't think of any large-scale solution that I've worked on in the past 5 years that used just one vendor. It's common to pull Microsoft, VMware, Altiris, Symantec, Citrix and various others into most enterprise projects these days. It's the nature of the beast. As Gabe mentioned, there is no one single solution for all - and honestly, we wouldn't be comfortable with having one vendor supplying every computing need, would we? lol In theory, it would make support easier, but you get back into that 'evil empire' syndrome and lose your ability to manage costs through competetive pricing.

Keep up the good work, Brian & Gabe - very much appreciated.

Thanks for sharing your thought process on this. It makes for a very interesting read. I think part of the confusion for some is that it appears after a quick scan that this has been implemented. A more careful read shows that this is a proposed solution, and not a final design. Perhaps a paragraph on the Proof of Concept plan (scope, test applications, user population etc...) would help avoid that confusion?

One more point -

As a University, this customer would qualify for educational pricing from Citrix. I am pretty sure that is at least a 35% discount of retail, and could possibly be more. I am not on the sales team, so I do not have the exact numbers in front of me. Here is a link with a description -

I am not certain if Ardence is under this program, however.

Further, Citrix recently joined the E-Rate Program. Here is a brief description of E-Rate -

"This eligibility means that schools and libraries can apply for funding and discounts to offset the cost of using Citrix Presentation Server as a part of their Internet and/or general network connectivity solution to deliver access to classrooms and other eligible locations."

So, not only can this customer get a 35% discount (or more), but they may also qualify for a subsidy. I think both of these factors should be included in your decision making process if both upfront cost and overall Return on the Investment are important. This info may not have changed your final math or your final decision, but at least all the available info would have been considered.

Just my two cents...

Barry Flanagan

Thanks Ruben (I saw your Ardence presentation at BriForum Chicago - you gave me a Grolsh)

Actually what I'm trying to do is not use Ardence. I had a Private image that I was streaming to a VM. I wanted to turn it from an Ardence Streamed VM to a flat VMDK file. VMWare Converter created the VMDK file sucessfully but everytime I reboot the VM it loses it domain membership. This VM is NOT being streamed its just a VMDK.
1) Brian - Great article. Thanks for sharing.
2) For all those "Guest" poster out there, your opinion is appreciated, but could you please state the facts to back up your opinion.
3) Believe it or not, IT is changing, and we do have to accept the changes. This is part of our jobs. IT is full of different solutions with problems. So, what is the big deal here?
4) Last thought, can anyone recommend a different solution other than Brian's solution. I love to hear it.

Anonymous Person.
"In the future, companies like CItrix and VMWare will have this solution tied up as one solution"

I agree exactly - this will be when VDI moves from a niche solution for hobbyists to the mainstream (like Terminal Server and Presentation Server are today)

BTW - My money is on Citrix to win this over the long run, even though VMware has more mindshare in "VDI" today.
This is a great topic, in spite of the skepticism expressed.

The previous post reflects reality as we've observered it:

"They have huge IT staffs running around fixing fires and no way to break the cycle. Anyone that does not belive that the model they put forth in this article, is not in tune with the changing business requirements customers are giving us today, is not in synch with the changing IT enviroment."

Unfortunately, it isn't easy for everyone to see this when the "current model" takes so much of our time. I agree with the previous post that the future is upon us and this isn't fringe stuff.

Brian, it is great that you are open to developing new solutions. Keep up the good work!
That's the job of integrators like HP and IBM. Their global support groups provide triage for customers. And that's what integrators do: integrate best of breed solutions.
CPS became mainstream way before they "tied up" all the peices - like RTO, Aurema, UP, Streaming etc.... Companies like VMware and Microsoft (platform guys) will provide base level functionality that innovators like Provision can augment.
The flaming is taking many forms not just the reaction of people not wanting to change the procedures/thinking. The number one issue that I deal with is people demanding solutions that are easy to manage and use. Many of the new solutions can’t really meet these criteria especially when it comes to long term management….”where will I be two years down the road” my answer is often totally muddy because of so many pieces in play.

So maybe the fragmentation of the industry is the core problem, not the engineers and architects putting the pieces together.
What’s funny is that same fragmentation looks good from other perspectives such as development and forward progress of technology.
People want the benefits of the multi vendor world but not the negative aspects. Take your cake and eat it but your going to choke on it.
1) 50 TS users on that hardware is the low end. There's a good chance you'll get 50% more than that (75 users).
2) 50 VM users on that hardware is the high end. There's a good chance you'll get 50% less than that (25 users).
3) I could just as easily say you'll get 3 times as many users (75 to 25) on TS desktops than you will VDI desktops and for half the cost of the necessary hardware.

In the end, the best solution would likely be a TS published desktop with all the TS compatible apps either streamed or installed. Then TS incompatible apps would be accessed in that same TS published desktop using Provision Networks ability to seamless publish VDI apps to the client. In this case, the client would be a TS published desktop. The cost savings with this is that the high cost of VDI desktops is only spent when a users need to launch a TS incompatible apps. All users would use TS desktops, and half the time those TS desktops would be accessing a VDI published app. Do the math and it's a much lower cost than giving every user a VDI desktop.
Dude - he already told you that Citrix doesn't have the core technology for the solution. E-Rate, subsidy, discounts, ebay... pigs don't fly...
I think you hit the nail on the head. Provision's solution offers the flexibility needed for this customer to deliver apps or desktops from any platform: TS, VDI, blades etc... publsihed or streamed via their integration with Softgrid.
Hi Brian,

Excellent Post - nothing like igniting the fires just before iForum? ;-)

On your and Gabe's behalf I take exception to those that are giving you guys a hard time on your *recommendations* to the University - especially those who feel free to shoot you down without offering any other alternative AND simply hiding behind "Guest"

It is great to see someone come out and be completly open about this process and bring it to the public domain where we can all have our 2cents worth - and for this you and Gabe have my thanks, as we're all trying to keep up with all the multitude of different options and potential solutions to help customers get the most out what they've got.

The domain membership is handles by Ardence itself.

Have you conducted an app review? Of the 1200 devices how many are using what apps and when? Is it worth the cost of implementing an environment that supports all 200 apps if 20% of them get used on 10% of the devices one day a month and some even less? Sounds like a great vision but they could slash the cost of this by taking a different approach; provide the anywhere devices with the most common apps and dedicated devices for the least used apps. Not as smooth but likely to cost less to implement and manage.
What is interesting is that 100 of the 200 applications are recognized as "problem apps" six months into the discovery process, and I have to read to this point in the discussion to find the first suggestion to review the applications. How many of the apps will be removed from the mix?

I sat in on a RES Powerfuse presentation this past week. The presentation referenced a terminal service published desktop environment, non-Citrix. The SE focused our attention on an icon within the published desktop. That icon represented an application installed locally on the client device. The locally installed app can be a Softgrid install, or standard install. The application is NOT installed to the TS, yet the icon and application function is available to the published desktop environment.

The RES Wisdom product controls chronological and dynamic installation of packages applied to workstations as well as server devices. While the Powerfuse product controls the ability of users to launch applications based on policies similar to the Citrix policy infrastructure, IP, user, group affilitation. Granular access to application resourses is also controlled (an incorporated Resource Hacker, with complete logging), certain users can access greater resource function within the app, while others have a reduced access to available resources within the application. Perhaps an opporutnity to reduce application count, while better managing resource access within the application environment? But I digress.

My take away from the presentation is that the Citrix Desktop Server can have locally installed applications "injected" into the desktop environment via RES Powerfuse. The workstation device as well as server build procedure is controlled via RES Wisdom. Yes, you will change it once with Ardence, but how? What complexity can be reduced even within the single image management requirement?

Scenario 1: Ardence delivers the image to the lab workstations, Softgrid bad applications and all. Six months out it appear to be working?

Scenario 2: Citrix Desktop Server has potential to deliver the "lab workstation" desktop, but is not required. (I have recently come off project where Citrix is negotiating costs for Desktop server licenses which reference a 70% reduction in license cost associated with CPS 4.0 CCU.) RES Powerfuse MSRP is $45 per workstation, and offers the ability to inject local applications installed to the client device into the published desktop environment (MSRP $270 per TS Server, I do not have RES Wisdom MSRP in my head at this hour). Interesting to note the Citrix Desktop server is not required for this service offering. Indeed, we were also shown the RDP desktop receiving the application "injection". CAD was referenced during the presentation.

Scenario 3: RES Powerfuse has the ability to deliver applications to the user client device without any Citrix involvement whatsoever. No WI, no PNA. An MS TS server delivering the app via RES client installed to the client device.

Scenario 4: Well, if we are going to eliminate bad applications, we should also eliminate bad scenarios. I have a hard time imaging any college student or professor that is not connected to the Internet 24-7. If these university folks can't find their way to the Internet, then I would consider this user activity as PTO from the network. If you want the apps, get online.

Finally, Citrix is pricing CPS out of market. They are raising prices to their largest and most loyal customers at rates that leave no alternative for business lines but to pursue savings in any direction to be found. Customers are getting pissed, and CIOs are telling engineers to be engineers again, and build the company something that delivers bottom line savings. If VMWare can do it, you better do it too application delivery guy.

Virtualization is all the rage, and TS represents the last bastion of one for one server computing in the enterprise. TS led the way on consolidation and savings, and now this leadership seems to have been overtaken, and overtaken substantially. The idea that a single vendor solution, which CPS has been for some time now, can be replaced by five separate vendors pitched by one of the leaders in the industry is a telling, troubling, and very exciting proposition. Good luck, and I hope to see the follow up on your development proposal.

I love all of the comments, both positive and negative. The negative comments are the same voices you will hear in the project meetings trying to find fault and resist change. Good to hear it here and be ready for the face to face meetings. Any project of this size will be complex, true, simple is better, but sometimes you have to decide who get the simple: you or the end users.

Besides, who wants Brian's web community to be a bunch of yes men (and women) ?

Good stuff..

Hi Brian,
The only big problems I see using Ardence is the driver support and reconfiguring all the gigabit switches. The vdisk of the common desktop would need all the newest drivers that cover 1000's of desktop hardware. The official way to do this is go to each different build hardware with a pci Ethernet card and capture the extra drivers for the vdisk image. if the hardware is very varied the could prove an impossible task not to mention making the possibility of making th OS unstable due to driver incompatibilities. Then you have to ensure the network is up to the traffic and that the switches are configured for Ardence.
I can see a good case for Ardence in the server room for PS vdisk provisioning but not on desktop hardware.

Hi Brian,

I think the solution above is a good one and will offer the customer a flexible and scalable solution and opens up as many opportunities to improve the infrastructure over other potential solutions as it opens up potential issues.
High (constant) availability for VMs
V Motion of VDI's across servers
Live backups
PXE booting stateless VDI VM's which will require less san storage
Total application isolation
The ability to oversubscribe hardware resources and prioritize according to rules

In terms of alternatives, to reduce the number of vendors if you replaced Ardence, softricity and VMWare ACE with Wyse's WSM provisioning product you could reduce the number of vendors, have a single application managing the OS images, Applications and offline streaming and improve the potential support issues raised above.

In terms of VDI brokers I'm not sure about the ability to publish applications but the Zeus broker is a comparable product to the Provision and VMware's (which I believe is still currently in beta) broker could also be considered depending on timeframes, as it would potentially reduce the number of vendors as well.

Also I think the VDI to server scaling figures above can be improved further using the new generation Dell and IBM diskless, virtualization dedicated servers (ESX installed on a flash card, more RAM, etc) making them a better proposition for this model as the more VM's you can get per box will reduce the ESX license cost as well as reduce the number of servers required.

Ugh, it's called VDI... You'll only need one image for the hypervisor.
I know that a campus agreement with MS allows more freedom to obtain paper desktop OS licenses, so perhaps for this environment it may not be as big an issue. But, in general, I rarely see the cost of obtaining a desktop Vista/XP license either addressed at all or properly when it is. It appears that properly licensing MS desktop OS's in a virtual environment is at least as expensive as TS, maybe even TS with PS.

Just looking for comments, not flames.
I think Thinstall is a great product but still needs to evolve some more. It's still learning to be an enterprise player but I do believe they (Thinstall) have the right long term vision. Hopefully I don't start a war with this comment but to be honest I still think SoftGrid is the most enterprise ready solution after finally having some time to play with SVS and Thinstall in depth. That being said the market space is still young and the competitors could catch up quickly with aggressive development schedules.
I think you are right about this. You would need to look at Microsoft VECD ( Vista Enerprise Centralized Desktop ) licensing rules. My understanding is that you can only license Windows for use in a "VDI" like environment (a) If you have a certain kind of Enterprise or Select Agreement with Microsoft and (b) pay an additional Subscription fee. Microsoft Desktop OS licenseing is very complex - if anyone out there can shed light on this - great.

PS: I think there is even a different license for Ardence PC ( like the 1200 in the Lab in this situation ) - not sure what its called, but think it is free to those volume license customers.
VECD information..
What if Microsoft were to leverage Citrix's acquisition of XenSource and provide Vista at a discounted price, assuming the customer were to implement VDI (DDI?) with Citrix/XenSource/Ardence? Make a deal with Citrix and allow the OS to be purchased at a discounted price (i.e., cheaper than the cost of the OS via VDI using VMware) and pretty soon, VMware's market cap starts the inevitable downward slope.

If VDI/DDI were to takeoff and Microsoft were to either acquire Citrix or make an OEM version of Vista for Citrix, the costs between VDI (via VMware) could be far greater than DDI (via Citrix).
Brian, how can you put 100 locall apps in one image. Will this image not be way to large... (cause it is about applications that can not be virtualized and they mostly are large and complex)
I can't help feeling the SoftGrid solution wasn't fully explored here. It sounds like the issues could have been resolved by hiring the correct resource for the job. I've been sequencing for over 18 months and was an application packager for ten years previous to switching to sequencing. Certainly, as with "legacy" packaging you do encounter applications that, initally, look like they won't work, however, with a little ingenuity and perserverance the issues can be resolved. I also do not agree that sequencing applications for IE is particularly messy with the correct approach. The reward for all the endeavour is not really in the virtualisation or application co-existence, for me the big win with SoftGrid is the provisioning. The ability to switch applications on or off per user through Active Directory is so worth the effort.

For me, this remit could be well served to a large degree with a well implemented SoftGrid solution, it just needs the correct skills applying to the job.


In my opinion; Citrix Presentation Server 4.5, and its various integrated services and functionality (ability to publish desktops, applications, content, isolated applications and application streaming) will handle many of the challenges you mentioned with compatability issues. Certain applications could also be packaged using "Thinstall" to further isolate certain applications that do not behave well together.

Citrix also offers the ability to package, publish and deploy the applications to all servers in the Citrix farm on a schedule. (This would save alot of money since the client would not have to purchase, license, support and manage deployment tools such as SMS, Netinstall, etc.)

Thinstall offers the ability to run certain application offline.

As you know; whatever solution, is determined to be the best for the university will require a best in class project plan, technical process, integration testing process, QA process and pro active system, application and server monitoring process.

Using Windows 2003 server and Active Directory, group policies could be used instead of flex profiles so you could easier manage all users as a group instead of setting each profile path individually.

And of course; with a great team, careful planning and a respectable budget, a solution can be developed and implemented to address almost any business and technical request.

Just some thoughts.

Garrick Sr.

I've not had much experience of Thinstall but from what I understand you need an ESD to provision the applications. From that perspective, any application that cannot be sequenced/profiled might as well be packed into an MSI and deployed via the ESD of choice. The virtualisation isn't much of a win if it's a handful of applications, which it would be on a properly implemented SG solution.

Is Thinstall really a player when it comes to large scale application delivery? From my knowledge it's only USP is that it stands alone as a self-contained unit, which, in my humble opinion, isn't much of a selling point.
How did you plan to handle streaming media (WMV, FLASH, MPEG, etc.) for new solution #2? (Published desktops: Use a VDI solution based on VMware VI3, using Provision Networks Virtual Access Suite as the connection broker and web interface. The VMs will boot from the same disk image as the lab workstations via Ardence.)
“dual-core $3,000 server can run 15 VMware VDI Windows XP instances, then today’s quad-core 8GB RAM can run 30. And for $6,000, you can buy an eight-core (two quad-cores), 16GB RAM, four 2.5” 15k SAS drives server which can run 50 simultaneous Windows XM VMs”

Some of those numbers are new to me and are very interesting. Could you please provide a little detail of how you would config the VM’s. How much RAM would be allocated to each VM and how far can you over-subscribe the RAM in this type of environment?
Project GAP: closing the gap.
The university brought in Brian and Gabe to advise them on a solution. Most companies do this and pay (I know it's hard to believe) folks like Brian/Gabe to provide third party input. If the customer could do this themselves, Brian/Gabe wouldn't be engaged with them on this project.

I would agree with your "experiment" comment if Brian/Gabe weren't the experts that they are, but that is not the case. Oh, and BTW, this is how most consultants make a living (shocking, I know).

THINK next time before you post!
Do you have any data how many Vista VDi VMs you could host per core?

As for as I understand the license, Microsoft require full Retail copy of XP or Vista unless you are using Vista VECD licensing. Am I correct here?
That was excellent, were glad to hear that you take the smaller vendors seriously too - without them, we wouldnt have been so successful with many of our deployments (or perhaps we could have tried to get around them with some workarounds which would have costs thousands in consulting and in the end still not have been a solid solution).

Many of the comments are also great...but its sad to see that some vendors are afraid of having competition, especially when their product for virtualization is just starting to get off the ground (can anyone say "Citrix"). Get off your high horse, and accept that you wont be a monopoly in the virtualization market.
I think Atari should have been added to the mix as well.
Of course it should have been added (doh!) - the admins will have lots of free time to enjoy it with a best of breed product mix!
Possibly Colecovision.
Isn't that what Microsoft has been doing for years?
Don't forget Commodore 64
Or as part of their severance package.

Think about it this way. Would any one of you want this solution to be under your realm of responsibility if you're on call 24/7?
If the Coleco Adam was part of this, then yes, I would love it to be part of my responsibility.
This is bigger than BriForum:

As soon as you require personalized PCs you have to have ardence differecning state, once you have this you have to patch each state, not the base image.

You can patch the base image but then you have to throw away the state.
So state has to be a layered approach. Ardence solves the problem at the base O/S level and it's included apps perhaps, but the personalization on top with other users apps is still an issue I agree. This is where I think the various app virtualization plays come in to play. Biggest issues I have with that technology to date I have is 1) it doesn't work for everything 2) You rely on a single service, that if fails kills all running apps, which for business critical implementations is a stupid architecture. Things like Thinstall are begining to solve this with agent less technology, but half their stuff doesn't work. So Ardence in summary, I see as an important enabling pieice of the technology. I think it will be a key part of the XenDesktop. Now they just need to get the app virtualization piece sorted out. Also really curious to see how VMWare plays this game as well. They have tons of missing pieces as well. Let's just hope that Citrix and VMWare figure out that the DDI/VDI market is huge and they need to work together and compete at the same time, as they have complementary solutions, as opposed to trying to hurt each othe.
I would like to recomend an interesting article from HP entitled "Desktop consolidation with HP Virtual Desktop Infrastructure using VMware Infrastructure 3"
One of the most interesting points is that a configuration based on the DL 380 G5, with 2 Dual core Xeon 5150 (2,67GHz), 16GB RAM and 2 direct attached MSA50 disk enclosures (20 disks for VMs) was able to sustain 85 to 100 Windows XP virtual machines (512MB RAM).
My regards for your great work.
I'm not sure what to make of the HP tests. At VMWorld 2007 the numbers we were hearing were more like 6-10 XP machines per core. These were also from actual case studies (see Northern Trust). Certainly a lot depends on the workload, but I doubt that anyone is going to approach 20 machines per core using VMWare. We are figuring on 6 and will be happy if it turns out to be more.
Just thought I'd offer a thought you may not have considered. I think a key consideration in the VDI world is how space the VMs take up. This affects the amount of time it takes to 1) provision new machines, 2) backup machines with user data, 3) how much time it takes to move the VM to new physical hardware (I think this will be really important when people start synching with laptops), 4) how long the application takes to startup the first time. Thinstall is the only solution that can function without caching to the local disk which is probably why it starts up several times faster than it's nearest competitor and it does increase the VM image disk size while the application is running.

> I've not had much experience of Thinstall but from what I understand you need an ESD to provision the applications.
> Is Thinstall really a player when it comes to large scale application delivery?

If you look at companies with more than 500 desktops, almost 100% have some sort of ESD in place SMS, Altiris, CA, etc... Application virtualization cannot be used to handle 100% of apps (doesn't handle device drivers, OS patches, etc). So you need an ESD for these things. If you have an ESD, why do want a complete seperate deployment system for a subset of your applications? If you look at what Thinstall and Microsoft are doing - both are moving to deploy virtual applications via .msi Thinstall already released this functionality while MS announced it will be available before 2008.

Why ? I think because a) people don't want two seperate systems, and b) softgrid servers don't scale to more than 10-20k users.

Softgrid servers are great if you can deploy 100% virtualized apps, but in a larger company this is not possible.

Also interesting on the packaging side Thinstall has partnered with Macrovision so you can build MSI or virtualized applications from the same user interface. I think this adds more data to the theory for a single system for virtual and non-virtual.

I predict Microsoft will eventually "deprechiate" Softgrid servers and they will copy Thinstall's model - it's already sounding that way with their announcing msi wrapper and release of a lighter weight server.
You know I'm really starting to think that Thinstall might be the app virtualization choice for VDI. Initially I was hung up on what it didn't do that I wish it did but really this technology really does allow for more agility in the VDI space. Just start launching apps and they will be streamed into memory. No more loading the global cache just get the VM online and start running applications.
I wonder how this project is progressing. Anyone know?
I guess no news is good news - much to the disappointment of that certain juice co. Makes me think that they've served up so much koolaid over the years, it's hard for the community to be objective anymore. Thanks Brian for having the rocks to think out of the box.
Citrix is in a shitload of trouble.
Did you say "koolaid on the rocks"? sounds like a drink they should serve at the next expo.
After talking to Brian in Antwerpen, i was reading this article, and most of the comments, and i have a solution for it.
I developed a version of SVS that runs in a terminal server environment. The product known as DVS4SBC has some great advantages.
Virtualized software talks to each other virtualized or not. It brings the best of Altiris inside Terminal server.
We can now package over 99 % of all software in a virtual layer, and activate and de-activate it at wish. Even a repair of a broken package is a single right click with the mouse, and then choose repair. In 5 seconds the applikation is working again as intended.
And best of all!
The packages can be used on Vista, XP and windows 2003. The same package for all three environments.


PS Brian, it was nice meeting you.
Hi Brian
Very good post. Just shows the endless possibilites within the Virtual Desktop arena. I have been trying to understand how if possible at all we could use Thin client terminals like Wyse etc along with this solution. Do you have any ideas?
You can absolutely. Provision Networks supports Wyse ThinOS terminals and all HP/Neoware terminals too.
So now we have three potential options for virtual desktop. VMWare, Citrix and Quest. Will Provisiing really grow from here. Quest has a very confusing strategy.

in response to your VDI requires too much management overhead, netapp has put forth a solution using FlexClones and a script. Check out this demo.

The interesting thing that I'm considering is that if this is all built on a single NFS Lun and I can spit out 100vm's in approximately 10 minutes combined with Provision's solution, I can take a minimal maintenance window for all of my master LUN's and patch/upgrade/etc., delete all of the Flexclones, and re-run my script re-creating my pools within minutes. If we can manage profiles and file-redirection properly and take away the user's ability to make changes to the desktops,there's no reason why this couldn't make a serious business case vs. traditional SBC.

This would make a killer large-scale deployment a breeze. Now if I can only get Netapp to write some GUI code around that script and tie it in as a plug-in to desktop brokers, etc.

Have you seen or heard Quest articulate their strategy?  If so, I would like to hear it too.


I just designed a spreadsheet to calculate the cost of VDI Solutions vs TS, based upon a factor of 4 VMs per server core.  With this I specified an IBM x3550 with Dual-Quad Core CPUs, 16GB of RAM, RAID Controller, 2 local 2.5" SAS Disks at RAID1 and a QLOGIC Dual-Port iSCSI HBA.  This configuration came to about $8000, before adding ESX or VirtualIron.  I did this with the impression that this system would comfortably support 32-40 VMs at 384-512MB RAM each.  I didn't go to 32GB of RAM, as this would be wasted unless I have real data that a CPU core can really handle >4-5 VMs per core.  Does someone have non-vendor data that a multi-core CPU can support more than 4-5 XP Pro VMs per core w/o sacrificing performance?


Brian, you've allready posted this article. Why recycling it now? Are you feeling the VDI wind in your face and didn't think anyone would notice?

brian, this is an old post. perhaps you want to modify it since xenapp is now free if you're streaming apps to xen desktops.
brian, this is an old post. perhaps you want to modify it since xenapp is now free if you're streaming apps to xen desktops. so you
brian, this is an old post. perhaps you want to modify it since xenapp is now free if you're streaming apps to xen desktops. so
Brian, can you comment on how you came up with 50 WinXP VMs on the server you described?  It looks like you are probably allocating only 256M RAM per VM, which could make performance poor.  We are running on a bigger server (same number of cores though) and allocate 768M RAM per VM and we get decent performance with 55 VM.  However, this is not a $6k server.

The university just asked Brian to propose a solution meeting all their criteria. Brian has done that with the best available technology. In June 2008, all their criteria can only be met by a multi-vendor solution. Its the starting point of the project, chances are they will decide which are the most pressing criteria and meet them first so as not to bite off more than they can chew. As per Brian's previous blog entry, by 2010, the situation may look very different. What is not in doubt is this is a really interesting blog entry (and judging by the number of responses, many agree with me, if not with Brian's suggested solution). In my experience in the UK, its the universities and colleges that are really interested in these solutions. Its fascinating to hear Brian's take on it. And as for whether the university appreciate this "experiment", are universities not supposed to be places of experimentation and learning? I say be bold, its what you Americans are famous for. Its why your universities are the best in the world, why all this technology comes out of the US and why we read Brian's blog!

Its nice how all of the people trying to bash ideas and start flames around here are all posting as guest instead of logging in. Anonimity can bring even the biggest coward out from the rocks long enough to throw a few... :-)
I was totally thrown for a bit untill i got to the bottom of this tread and realized how old it was. If you applied what citrix is doing with Xendesktop Platinum to this scenario, I think it would play out differently. In addition with what citrix is working on with its new isloation enviroments providing interoperability between isolation enviroments I think things would be different


how did this case study end ?

and do you still feel the same for this case looking at all the new features in vmware,citrix and of course softgrid.