There are three fundamental difficulties facing any hosted desktop solution. They are :
1. What to do with the desktops that can’t be virtualised?
2. What to do with the desktops that can’t be virtualised?
and, most importantly,
3. What to do with the desktops that can’t be virtualised?
Of all the vendors in the hosted desktop space, Citrix has been delivering desktop virtualisation solutions the longest. As such, perhaps they are the most aware that an enterprise desktop strategy isn’t about delivering a single solution. A solution needs to be flexible enough to present a variety of services to a range of devices. This isn’t just about having different client support, but about delivering applications and data either to different environments: secure and insecure, managed and unmanaged, fat and thin.
Perhaps the most important aspect of Citrix’s product portfolio is its FlexCast model. Other hosted desktop vendors have a similar option. Some do not.
What is FlexCast? Why is it important to customers? Can a hosted desktop vendor survive without having something similar?
What is Flexcast
Different types of workers across an enterprise need different types of workspaces delivered to different devices. Some require simplicity and standardization, others require high performance and personalization. Importantly, users may need to change between types of workspaces depending on their activity.
Citrix FlexCast isn’t a product, or a protocol – but rather the concept that IT should be able to deliver a variety of types of virtual desktop with each tailored to meet the performance, security and flexibility requirements. There is no one size that fits all so why try and shoehorn users into a particular solution?
Flexcast has five delivery options:
- FlexCast for Apps: Hosted Shared, Hosted VM, and Streamed: The goal for Apps is allow Microsoft Windows applications to be centralized and managed in the datacenter, and delivered as a service to physical and virtual desktops. By either using the published application feature of XenApp (or XenDesktop) applications are delivered via ICA for use while connected, or you can opt to use application virtualisation (with either Citrix’s own solution, or Microsoft App-V) directly to the endpoint for use when offline.
- Local Virtual Machine: Using Citrix’s client hypervisor solution XenClient, local VM desktops can offer the benefits of centralized, single-instance management to mobile workers that need to use their laptops offline. When they are able to connect to a suitable network, changes to the OS, apps and user data are automatically synchronized with the datacenter.
- Streamed VHD Streamed VHDs leverage the local processing power of rich clients, while providing centralized single-image management of the desktop. It can be ideal for environments that want to use diskless PCs for maximum data security. However, while it does offer centralisation, there is a requirement to have a robust networking and delivery infrastructure to maintain the delivery of images to end devices.
- Hosted Desktops Hosted desktops offer a personalized Windows desktop experience, which can be delivered over the network to a wide ranges of devices. Hosted desktops are persistent – the user will connect to the same virtual machine each time, every time. This option combines the benefits of central management with full user personalization.
- Hosted Shared Hosted Shared desktops provide a pre-configured, streamlined and standardized environment with a core set of applications. Citrix offer two options here, shared desktops can be provided either with XenDesktop, or with XenApp. XenDesktop offers both an x32 and an x64 desktop OS environment whereas greater density of concurrent users per server can be gained with XenApp – all be it only on an x64 server OS.
Arguably, there should be a sixth component: managed end-point without virtualisation. Hosted Apps deliver the apps – but not the OS. While there is Local VM option (and I’m a big fan of Client Hypervisors), it is not viable for all devices. Sure, VHD Streamed is one solution for OS delivery, but there is no option within the Citrix FlexCast model to deliver, manage and maintain a per-desktop-OS: this is still an important model for many organisations. A cheeky acquisition of an innovative company like Wanova could plug that particular gap, and would Microsoft be that bothered and see it as a challenge? Unlikely. It would offer an alternative to the “VDI” tax.
Importantly, no part of the FlexCast model requires a secondary/different Citrix license. True, there is a difficulty from an administrative management model that the different model types don’t have the same management interface: but this is road mapped to change over time and is unlikely to significantly impact on the Capex for a project
Flexcast Alternatives
Microsoft? All the VDI vendors are in fact offering a virtualisation solution to host applications designed to run on Windows. To run through the FlexCast model Microsoft has Remote Desktop Services for hosted desktops and hosted shared; application virtualisation with App-V; delivery of local VMs using Med-V or Windows XP mode. And of course, installation services allow you to deliver the OS to devices on your network.
Quest’s collaboration with MokaFive allowed Quest to quietly talk about the “Virtual Desktop Continuum” – with MokaFive’s Live Player adding Local VMs to compliment Quest’s PV/VDI solutions. Quest & Ericom don’t have their own desktop virtualisation solutions: but here they could argue they both integrate access to and delivery of Microsoft’s App-V.
| Product | Presentation Virtualisation/Terminal Services | Virtual Desktop Infrastructure |
Application Virtualisation | Local Virtual Machines | Streamed OS Delivery | Traditional Desktop Management |
| Citrix |  |
|
|
|
|
|
| Ericom | |
|
||||
| Quest/MokaFive | |
|
|
|||
| Microsoft | |
|
|
|
|
|
| VMware | |
|
VMware’s VDI failing, a one trick pony?
An issue with View is that a VDI-only solution is a limiting strategy when deployed on its own. For some use cases (off-line access) it is irrelevant, for others (high numbers of task based workers) it is expensive. As long as VMware is only in the VDI niche, Citrix is in a commanding position to keep dominating VMware in the market – while Citrix owns that circus, View could be considered just a one trick pony.
VMware need to drop the idea that desktops are only about VDI and embrace a wider delivery platform: however, VMware is reducing its delivery options – their local VM option ACE is OA as of December 2011. Project Horizon is important because it gives VMware a chance to offers organisations the option of breaking out of the requirement to deliver a desktop and focus on the applications. But of course when Horizon ships there is and will be a substantial gap between “what is in the workspace now” and “a pure SaaS delivery mechanism”. Deploying virtualised application packages is insufficient: this is application delivery, but you also need to consider data delivery. For many applications for businesses today it is the marriage of the data and application together that needs to be virtualised and presented to users for those users to work effectively.
Perhaps an acquisition of the likes of Ericom could offer an opportunity for VMware. Ericom’s PowerTerm WebConnect provides a management interface for PV, and they have developed an innovative remoting protocol in Blaze that not only offers an alternative to RDP and could compliment PCoIP as well as being one of the first vendors to deliver a usable HTML 5 client offering.
Flexibility is Key for Workspace Delivery
Can a hosted desktop vendor survive without having something similar? Of course. But, can a customer offer a viable and cost effective workspace delivery platform if they only use VDI?
If you are able to utilise Presentation virtualisation/Remote Desktop Session host whenever you can you can reduce the cost per user significantly over VDI. Having a range of flavours better suits user’s tastes. Local VMs, client hypervisors should not be considered as simply “Offline VDI”. If you have a device capable of delivering a quality experience with a local VM then why not use it all the time and avoid the need for additional costly datacenter hardware running hosted VDI for each user. And of course, do not ignore the question – “what to do with the desktops that can’t be virtualised”.
Good article, but I don’t think you’re being realistic with Microsoft’s offerings to rate them in the same league as Citrix.
Take med-v as an example, its not even remotely comparable to XenDesktop when it comes to features and flexibility. Similarly Windows XP mode / virtual pc are at best hacks when compared to the capabilities of xenclient.
Even the delivery method of HDX vs RDP are miles apart in features.
Personally I’d make the microsoft offerings yellow ticks, to indicate a product but not as feature rich.
It’s great to see the other vendors offerings too, thanks for the great article.
Valid points Andrew. It wasn’t an intention in this article to do a feature by feature call off: my other reader often highlights I have enough trouble keeping him awake beyond 500 words. It would make a good follow-up piece for sure.
It is also fair to say that Citrix.. and Ericom & Quest offer services that enhance the core services given by MS. “not in the same league”? Well, its a canny starter for a title.
Remember HDX & RDP are two very different things. HDX is a set of technologies, not a protocol. If it were apples vs something-more-similar-to-apples then it would be ICA vs RDP: and then I think it can ends up being moot depending on your circumstances. Would those feature differences be important for a LAN based environment?
Med-V isn’t XenClient that is true although they do both support local VMs . Different features, different things. But then if all my PCs have AMD chips sets I’m not going to be able to use XenClient. And if I don’t have XenDesktop and I’ve can’t use XC in an enterprise.
Ultimately, Citrix aren’t alone in offering a range of services: there are other options. Importantly, (I think) there are still gaps in the flexcast net. From VMware’s perspective its less of a net, and more of a single line.
Thanks again for the feedback – definitely will revisit and expand on this I think.
@ Andrew (Morgan)
I’m in complete agreement. Technically MS can have a tick in the box, but the limitations start to show very quickly once you try to deploy in anything but the most benign environments.
@ Andrew (Wood)
There is another aspect to Flexcast that’s deserved some attention, Flexcast offers a degree of context awareness so that an application can be streamed or delivered using presentation virtualization depending on the capabilities of the endpoint, rather than requiring specific action by the end-user. Citrix has a long way to go to fully develop context awareness, but it has at least started down this path. So far none of the other vendors offer even the most rudimentary capability here. However, VMware is clearly heading down this path with Horizon and hopefully will deliver a meaningful implementation at some point in 2012.
Simon
@Simon – “technical limitations start to show very quickly”. I think since 2008R2, with win7 that the “very quickly” (which was most definitely a case in windows 2003/2008) is taking longer. It’d be an interesting thing to explore for sure.
I’d also say there’d be benefit in flexcastesque solutions having a more flexible concurrent license model. Here MS obviously shoot themselves in both feet with both barrels; but even Citrix charges a larger premium for this than (say) Quest or Ericom.
It is true that that there is a level of interchangeability with Citrix’s streamed applications/published apps that is unique. I think there are at least three considerations :
a) it still doesn’t address the delivery of non-virtualised end-points/applications. I know many would love not to care about them, but they’re there.
b) having context awareness delivering the application locally is great, but the data needs to go too. Its no good if angry birds is delivered if you don’t also send the scores and completed levels. Here yes, Citrix have some great potential
c) it relies on you deploying an agent on the end-point
Citrix have some whizzy features in the Receiver: with more coming tied into their Storefront technology. However, how will context awareness operate in a multi-landlord solution? (which is a term I’ve just made up, where you are the recipient-of-many-services rather than the owner-of-a-service-to-many-customers).
> a) it still doesn’t address the delivery of non-virtualised end-points/applications. I know many would love not to care about them, but they’re there.
Agreed, Citrix has shown remarkable reluctance to acknowledge the continued presence of non-virtualised end-points/applications (something that I’ve been telling them for years). I think that it is now too late to expect it to come around to this perspective.
> Its no good if angry birds is delivered if you don’t also send the scores and completed levels.
That’s a perfect example of just how incomplete today’s “post-PC” workplace is. For all the vendor hype, no one yet offers a fully integrated platform that synchronizes applications, data and context between platforms. Depending on how broadly you define workplace, both the app problem and data problem are progressing towards viable solutions (even if they exist more as checkpoints on a roadmap than as released products), but there’s no single integrated solution that ties application and data together yet, never mind the far more difficult challenge of sharing context across multiple devices. It’s something I’ve explored in more depth on my personal blog here and offered the perspective that the most complete post-PC implementation today is the Amazon Kindle which as you rightly point out is very limited in what it can offer.
As the physical point of access into the workplace becomes increasingly diverse, the problem of synchronizing applications and data across all only get worse, however it does look as if Citrix and VMware are on track to deliver integrated solutions although I suspect it will be several years before either offers a complete solution and I have concerns about the approaches that both companies are taking.
Indeed. I believe the “Angry Birds Scores Conundrum” isn’t fundamentally a “PC” problem, or a “workspace” problem – its an application developer/seller problem.
Traditionally application developers will have kept their data close to the application, utilising OS calls that saved data. Typically those calls were to the local OS (saved to part of the file system), or perhaps if you were really funky to a database (but you then had to own a database)
Ideally new applications/application updates start thinking outside of that standard pc-box and write their code to sit across platforms: perhaps even offering the option of hosting data storage services themselves.
But then you can get into issues about who owns the data, manages the data, recovers the data, secures the data. More importantly, it doesn’t help with the huge array of “legacy” applications that are available and being actively used today.
Application sync is relatively straightforward I’d suggest – but the data.. not only the application configuration information but most importantly the user data that is a huge can of worms.
There are some nice data sync options coming out. Citrix’s Sharefiles has some useful options with regards to facilities to offer drive mappings and security but that’s only a small step on a very long road.