Brian Madden TV #20 - XenDesktop 4 license changes, plus our final VMworld 2009 wrap-up

In this week's Brian Madden TV, we start with a conversation about the community's reaction to Citrix's decision to change XenDesktop licensing from concurrent user to named user.

Then we'll move on to our final segments from VMware's VMworld 2009 which took place last month in San Francisco. Brian got a chance to talk to Microsoft's general manager for virtualization, Mike Neil, about Microsoft's VDI bundle.

Finally, we'll listen to folks from VirtualStorm's explain how their VDI solution had 452 VMs running on one blade.

View All Videos

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Fyi, total XA licenses sold = ~20m; total XA users = ~100m.  Your comment that there have been about 100m XA licesnses sold in total would imply revenue to CTXS of $25B (assuming a $250 ASP), which is not possible.  The total historical revenue from XA licenses in more on the order of $5B.


I just want to comment on the application virtualization part, you mix it up with Citrix. The applications in a VirtualStorm desktop are running in the virtual desktop machine. They are not offloaded and displayed like a Citrix application is. Your running application is using your virtual cpu and memory resources, just like it would on a normal desktop. There are no tricks there! The only strange part is the cross machine shared disk space. Yes, space! Not an application server, nothing running there, it’s just disk space.

There are only Windows XP desktops in the whole environment, no terminal servers, no application servers or what so ever.

I’m one of the users of VirtualStorm and what I really like is that they add virtual applications to a virtual desktop environment in the most efficient way. Nothing old-style per client download over the network… just add it once to your shared disk. Plain and simple!


What seems that VirtualStorm is doing is segmenting Windows up into What Changes, What doesnt Change, What needs Speed and What doesnt need Speed. Then they have solutions that optimise each of those.

They probably have some profile managment for data that changes.

They have the base VM image it self what what doesnt change.

They have the offloading of the pagefile to cache, and probable alot of deduplication going on there.

They then use the memory sharing feature of ESX.

Even though not all of those VM's were doing much if at all then it still it still reprisents a real world situation in a way. Say you have 1000 employes, each with their own virtual desktop, but you only have 250 working at the same time. What VirtualStorm would able you to do is have all those 1000 desktops ready when they are needed.

So basicly what they have done is create a vdi solution with a ts density.

Very cool :)


I'll put in my $.02 here.  I am the lead architect from GreenPages who designed and built the infrastructure supporting the technologies shown in our booth at VMWorld 2009.  

We did in fact have 2650 VMs booted and running across 6 of the Cisco UCS blades (the other 2 blades were used for running infrastructure VMs).  No funny business on the back end, the gear you saw on the show floor was all we had.  Each blade was configured with 2 quad-core Nehalem Intel CPUs and 48GB of RAM.

We actually had to build the environment from scratch beginning on the Friday morning before the show.  In fact, up until we arrived in San Francisco, the blades, Nexus switch, and EMC array had never been in the same location nor connected in any way.  The only pre-staging that was done was the creation of the managing infrastructure VMs (Domain Controller, File Server, Virtual Center Server, etc.) along with the Virtual Storm core 1.2GB base image and the application repository of approx. 90GB.  It was very challenging to say the least to get the entire environment spun up in such a short amount of time.  

All this being said, we are not in any way advocating that running 480 VDI sessions on a single blade server with this configuration is a production ready solution.  However, densities approaching 200 VMs per blade are very possible in a production environment.  The point of the booth was simply to demonstrate how far we could scale a VDI environment using technologies from our key partners (VMware, Cisco, EMC, and Virtual Storm).  

In fact, by the end of the show we were trying to break the environment and did succeed in doing so.  It turns out that Virtual Center was the first piece to break as it simply could not handle managing close to 3000 VMs.

GreenPages will be hosting an event at Gillette Stadium on November 12th where I will be presenting on the architecture of the booth.  I'm also happy to respond to any questions from the community regarding technical aspects of the configuration.


OK, so what do the guys from VirtualStorm have to say for themselves? First of all, it’s good to see some folks in the community getting part or all of the picture. Second of all, we’ve tried to make our technology as clear and transparent as possible, which is why our design philosophies are there for all to see on

Like we explained to Brian and a number of folks at VMworld, any technology that you wish to employ to solve a specific problem, must take into account the whole of the environment in which you deploy your solution. That said, the biggest obstacle to any VDI deployment is actually the environment, which was never designed for large scale centralized virtualized desktops and hasn’t been since the time of the mainframes. Sure we can try to hack away at some of the issues, but unless we take a step back and look at the situation as a whole, we’re not going to solve these issues without finding new ones.

We took that step back. We designed on requirements from our customers and relentlessly pursued one single goal: great end-user experience. This required a new way of thinking about desktops and their virtualization, of applications and their virtualization, of desktop and application deployment and automation in a complex, centralized IT environment. It also required a perfect balance between the desktops, the infrastructure, the virtualization and the management of the desktops and their applications in order to scale to thousands of desktops in a single environment.

In the end, most of VirtualStorm is about properly designing your (desktop) IT environment. The special pieces we built were a Windows Memory Enhancement Stack, a Windows Disk IO driver, an agent to manage the application virtualization (DVS4VDI) and an intuitive Management Console that knows how to scale. This all in an open and easy to integrate format that allows you to connect any end-point, use any broker and to automate most of the drudgework.

The result is a centralized virtual desktop environment that reaches extreme VM density on standard server hardware and good, but not excessively expensive, SAN storage. It’s fast in deployment of desktops and applications, can automate most of the regular tasks, allows reduction in most of the costs associated with desktops and management of desktops.

But seeing is believing. We showed off on Vmworld with 460 VMs on a single blade (in real life we recommend 200 to 250 of these VMs). But we did so without trickery or ‘magic’. Anything that’s advanced enough will look like magic to people who don’t understand the technology. But there’s a simple way to find out: just ask and learn.


Yaaaaaaaaaaaa,its a very nice artilce about this topic Brian Madden TV #20 - XenDesktop 4 license changes, plus our final VMworld 2009 wrap-up ,i read this article in this artilce have more inforamtions.Thanks to share your thought wihtus.

<a href="">Forex Trading System</a>