The Value of VDI Trade-Ups: XenApp to XenDesktop or VMWare View, Is It Worth It?

Which is better – Virtual Desktop Infrastructure or Presentation Virtualization? If you have, say, a Citrix XenApp PV solution – you may may be tempted at the reduced cost per license of VDI – XenDesktop appears to be almost half the cost per user than XenApp. Maybe you’ve a VMware vSphere back-end and the thought of having one platform to manage is appealing. Maybe you have a different solution – perhaps Quest’s vWorkspace or Ericom’s Webconnect and are wondering what all this fuss about moving between PV and VDI is all about?


Is there value in trading your PV licenses to convert them to a VDI?
There are situations where converting from Presentation Virtualization to a Virtualized Desktop Infrastructure can be beneficial: for example, you need to give users greater autonomy within their session; new applications need to be deployed that aren’t supported on a Windows server environment; you can no longer support Windows 2003/8 x32 and must have x64… but your applications don’t work on that OS. All valid reasons, but just converting for the perceived gain of Capital Expenditure is a bad approach.  If a Presentation Virtualization environment is efficiently delivering applications (and desktops for that matter) then the fork-lift upgrade to a VDI solution can prove to be have a greater operational expense in the end.

Why? Because :-

  • VDI is not ‘server virtualization, but for desktops’
  • It is not simply the broker licenses that change.
  • Your Licensing count or costing may alter
  • What is the change in your user experience by moving to an alternative remote protocol?
  • What is the impact on your administration?

There is an interesting underlying question with a ‘trade-up’ question. Citrix have two products – XenApp, and XenDesktop. XenApp allows you to have some VDI functionality; XenDesktop allows you to have VDI and publish applications with XenApp. Effectively very similar functionality yet they are priced, and licensed differently: XenApp is the more costly, and is licensed per concurrent user. XenDesktop is cheaper – but typically is priced per named user, or per desktop (although there is a concurrent option with some caveats). VMware has View – a VDI product that is licensed per concurrent virtual desktop. In the main – the  “trade-up” question is often ‘do I move from XenApp – which has an expensive per concurrent user cost to the seemingly less expensive View or XenDesktop?’

Bear in mind there are a number of alternative solutions to Citrix and VMware – from vendors such as Ericom and Quest – that have both PV & VDI solutions licensed and managed as one product: typically for a reduced license cost for Citrix or VMware. What happens if you are a Citrix XenApp customer and have never heard of Quest’s PV solution vWorkspace? Well, you can read more in our freely available whitepaper here – and of course, read on. However, even then saving money on license costs should not be your focus as any move involves much more than simply ‘changing licenses’.

VDI is not ‘server virtualisation,  but for desktops’
VDI can be complicated. VDI can be expensive. Why? Because a successful VDI implementation necessitates good management. You can be lax with the management of traditional desktops, (indeed this often leads to a high cost for management, which prompts you to look at alternatives) not so with VDI. Disk, Image and Profile Management are three areas that come to the fore significantly in a VDI solution.  The cost for managing a desktop does not come down simply by introducing VDI – if your VDI is poorly managed, desktop management costs will still be high: higher possibly given the data-centre infrastructure that VDI requires.

Traditional desktops often exist as assets that aren’t optimally managed –  and the smaller the company is, the less standard approaches are used. Move that poorly managed environment into the data-centre and the resources required to host those images will be costly to install and maintain and provide your user’s access in a manner to which they are accustomed. Even when you deploy tools that can provide a VDI environment ‘on demand’, such as Panologic’s Pano Express or Kaviza’s VDI-in-a-Box, there needs to be a consideration of how is that ecosystem – the servers, the network, the  client devices managed?

With server virtualization – you can consolidate your servers onto fewer devices. With VDI – you host many devices on servers in the data-centre – but you still need to manage and maintain the end devices: even the ‘zero client’ devices have to be updated: and so, managed.

In moving away from your PV service, you need to consider that  your management needs to change. Now it is not one OS, and one application set per server – but many OSes and their application sets. How do you update those applications across those images? You may used to have had a very simple thin client environment. The device  booted, it ran a basic client – there was little need to manage the device. This may no longer be the case with VDI. Maybe you have to change all your devices to support new protocols (such as PCoIP); or new functionality (such as Citrix HDX, or Quest EOP, or Ericom Blaze) – in which do your existing clients need to change, and if they do how are these new devices managed?

What is the impact of Re-licensing?
In evaluating a PV->VDI program, often a significant missing piece is “where are the costs of the associated re-licensing”? VMware View includes ESX licensing (there is no extra cost over that of the View licenses). Obviously, Citrix XenDesktop ,or Quest vWorkspace et al do not have a built in ESX license: implementing those solutions will therefore incur additional cost if selected and you had decided to use vSphere environment.

That said, you can host XenDesktops desktops on XenServer: for the Enterprise and Platinum editions of XenDesktop you are allowed to use XenServer Enterprise edition for hosting XenDesktops. You could also use Microsoft’s Hyper-V R2 – which is freely available; but that hypervisor currently does not support advanced features such as memory expansion.

Regardless, your VDI solution is likely to require a hypervisor management cost that is going to add to the operating costs when you’ve migrated from PV to VDI.

In addition, do you have the correct end device license? If you’ve a PV solution you will have a Terminal Server CAL or Remote Desktop Services CAL. VDI licensing for Microsoft is different although as of July 1st 2010 Microsoft licensing became easier for VDI:

  • Windows Virtual Enterprise Centralized Desktop (Windows VECD) and Windows VECD for Software Assurance (SA) no longer appear on the price list – they have ceased to be.
  • Virtual desktop access rights has become a Windows Client Software Assurance benefit. If you intend to use your PCs covered under SA, you’ll now be able to access your Virtual Desktop Infrastructure (VDI) desktops at no additional charge. Ideally, Microsoft would like you to move to SA.
  • Interestingly, if you want to use devices such as thin clients, or other devices such as smartphones, that do not qualify for Windows Client SA you need to license those devices with a new license called Windows Virtual Desktop Access (Windows VDA) to be able to access a Windows VDI desktop. Windows VDA is also applicable to third party devices, such as contractor or employee-owned PCs.

Windows VDA is a device-based subscription that has been designed to help organizations license devices that do not qualify for Windows Client SA (such as thin clients, contractor PCs, etc) to be able to access a virtual desktop. Windows VDA is available for $100/year/device through all major Microsoft Volume Licensing (VL) programs.

You may well have invested in RDS CALS – this is a different instance from VDI. Do you now need to re-license? Bear in mind an RDS CAL is a one-off cost. SA and VDA are a yearly expenditure.

Impact on users per server
For concurrency, VDI or PV – which is better?  Here we’ll defer to the excellent work done by the boffins at Project VRC who have spent a great deal of time and effort comparing different PV/VDI solutions on different hardware platforms. (Which in my opinion (i.e. me – Andrew Wood) should be more widely referenced – and it is a great shame that they’ve not been  re-invited to present at VMworld this year). Their findings show that, ultimately, a PV solution provides a greater density of users (i.e. users/server) than VDI on similar hardware. Read their reports – which are detailed and informative and a must read for any virtualization architect.

However, the likelihood is, for the servers you have at present hosting PV, deploying VDI on the same hardware will mean fewer concurrent users per server. If you want this to change, there will be a need to upgrade or add-to your server farm: which is an additional cost.

User Workspace Management is a priority
User Workspace management is a problem in PV. When you have many PV servers (which is very likely) : you’ve likely worked hard to ensure that user settings are consistent regardless of the server the user logs into? How to you persist their user profile settings, their mapped drives, their printers, their desktop view between servers, between silos? They need to be the same otherwise you generate unnecessary support calls; the user is unproductive.

Guess what: this is no different in VDI.

Now, a subset of user workspace management is “profile” management i.e. how quickly does the user profile (i.e. the user’s application settings)  load and unload?

VMware acquired RTO Software’s profile optimization tool. This piece will have to wait until the next release of VMware View to be available. Although there are reports that the next release is delayed, we understand VMware View 4.5 is on track and on schedule. However – what is important to note here is RTO’s profile software (which is to be incorporated into 4.5) was focused on delivering profiles quickly – but not on the management of those profiles. Citrix bought into Sepago for their profile management solution – and this solution is built-in to XenApp and XenDesktop (so no license change there); but again that solution is focused on improving logon times, not managing the workspace.

To best manage the workspace you really needs a solution such as AppSense‘s Environment Manager, RES Software‘s Powerfuse, Tricerat Simplifysuite, Liquidware Labs or Scense. What tools are in your PV environment that are being used here? If you’ve got tools, does their licensing need to change?  How is this cost being incorporated into your program?

What is the change in your user experience by changing the remote protocol?
With the release of the PCoIP client VMware is making strides to match up with Citrix’s HDX (me ,Joe Jessen believes with the introduction of PCoIP View has leapfrogged ICA on the LAN but I concur on the WAN Front and lack of device support) (me Andrew Wood thinks as its more often about the WAN and access over the Internet – PCoIP struggles here – ideally an organization utilizes a single protocol)

But if you’re moving from PV – especially if you’ve XenApp – there is still a gap in supporting multiple client device types, especially mobile devices, if you chose an RDP based solution.  VMware’s answer to poor PCoIP performance is to have the users drop back down to a straight RDP client.  Quest have EOP as an ‘enhanced RDP protocol’, Ericom the same with Blaze. Yet, both require a specific client install. Will your trade-up process exclude devices that your users rely on for connecting to your organizations applications?

Staying with a Citrix solution would at least not change the clients the users have to access their XenAPP environment today.  Although that appears not a huge expense, there is the risk of higher operational expenditure in supporting an entirely different client or access method.

What is the Impact on your Administration?
They’re just admins – what does it matter? They’re geeks – they’ll love a challenge. Maybe. But more likely, maybe not. You’re changing a business process here. New employee, new job (which it is) so – 3-6 months before they are ‘operational’ if we’re being being honest. Have you costed in re-training of your admins?

Is there value in a VDI Trade-up?
At VP we believe that an organization’s desktop strategy should look to include the best desktop options – not a desktop option. That might include one or many of physical desktops, presentation virtualization and virtual desktops.

Perhaps you are looking at your XenApp renewal in the current economic climate and thinking ‘can it be cheaper’? As we’ve mentioned, a function of XenDesktop is to allow published applications from XenApp – so if you’ve clients that support published applications – you could convert and lose almost no functionality: you couldn’t run full desktop provision only published applications. XenDesktop is of a similar price to View: does it offers similar functionality? At the moment, in comparison to View, XenDesktop’s extra benefits include a profile management tool, wider client support and a remote protocol that provides a good user experience over a WAN. Here may be then a benefit of ‘VDI trade-up’ – the infrastructure doesn’t change, but the license type does. However – with this type of ‘trade-up’ be conscious of the fact that XenDesktop’s licensing is different to XenApp – does the move away from a concurrent license model have impact? And, the management and infrastructure requirements are different.

Should you move from PV to VDI? No. No unless your requirements have changed from what led you to introduce PV. If however it is ‘users would like to see Windows 7 interface’ there is a business case for maintaining the status-quo – you’ve already the environment, you’ve greater concurrency, there is no migration or retraining costs.  Licenses may be cheaper – but licenses are not the only cost.

When is it ‘yes’?
It is ‘yes’ when requirements change. VDI enables you to deliver desktops to users who need the greater flexibility of a personal workspace and/or a desktop OS. Now here there is a benefit in managing both a VDI and a PV environment as a single entity – bear in mind two major virtualization vendors  are not capable of this – neither Citrix or VMware can accommodate a single instance management console for PV & VDI: for this you’d need to select a vendor such as Ericom or Quest.

One reply on “The Value of VDI Trade-Ups: XenApp to XenDesktop or VMWare View, Is It Worth It?”

  1. I’m glad you mentioned the option of having a mixed environment of Presentation Virtualization and VDI. Too many people in IT make this an either/or choice.

    The fact is that Presentation Virtualization has advantages over VDI, while VDI also has advantages over Presentation Virtualization. I’d agree that most organizations are best served by adopting a hybrid approach, with an optimal mix of Presentation Virtualization (for
    task-oriented users), VDI (for power users), and Blade PCs (stock traders, graphic designers, etc.) which delivers the most benefit and platform flexibility to the organization.

    Adam

Comments are closed.