VMware is going to do Linux VDI too. So what if it’s a niche of a niche, people need it.

VMware recently announced that they will be building Linux support into Horizon View.

A flare-up on twitter yesterday around Linux and VDI between AppDetective (@AppDetective) and Mark Lockwood (@_mlockwood) got me thinking a little bit more about the idea. Their conversation started because VMware recently announced that they will be building Linux support into Horizon View, which is not usually the kind of thing that gets old-school Windows guys-turned-bloggers all worked up. It is, however, something that the industry can support. For years the likes of Red Hat, Virtual Bridges, and NoMachine have been selling solutions based on delivering Linux VDI. There’s a market even though it’s a niche of a niche.

In August, Citrix released a tech preview of Linux Virtual Apps and Desktops delivered from XenDesktop and XenApp, which conjured up memories of my one foray into MetaFrame for Unix way back in ought-two. In that project, they had an IDE that was running on Solaris and weren’t getting the performance they wanted through X11 sessions. They had to use Unix because they had an app that required it, and so they needed an answer to their problem. Was MetaFrame for Unix mainstream? No. Was it helpful in specific situations? Yes.

Of course that product is no longer around, and I doubt we’ll see a Unix-specific desktop virtualization solution ever again, but the spirit lives on with these platforms that support Linux. There are two types of potential Linux desktop virtualization proponents: the cheapskates and the people with real business needs. The cheapskates look to Linux because of favorable licensing, and the most ambitious of them can even do it for free. Windows licensing is expensive and that’s a mighty big line item to remove from your IT or VDI budget. The people with real business need are those that are running Linux already simply because they have to. Only one of these types of people will be successful.

If you’re solely thinking about moving to Linux to avoid paying Microsoft any licensing dollars, boy do you have your work cut out for you. Odds are you still have Windows apps, and to support Windows apps you need…Windows. (Don’t try to sell me on Crossover, it’s a non-starter in most enterprise scenarios.) So if you’re trying to move to Linux VDI just to get away from Windows licensing, you’re also going to have to move your apps. Good luck with that. If you don't have any Windows apps and you're switching to Linux just to save licensing, do you even need a desktop? (And who are you, anyway? You lucky, lucky person.)

Those with real business needs for Linux applications aren’t concerned about cost. They’re concerned about meeting their goals, be it more reliable GUI remoting, accessing apps from anywhere, no data on the client device, or any other old-school benefit of desktop virtualization. These are things that probably can be done today with various tools, but require more work to implement and support than they’re worth. Packaging up desktop virtualization solutions for these users is just what they’ve been waiting for.

Citrix and VMware are entering the space not because they think it will help the community. They’re doing it because they have customers asking for it. Even though Linux desktop virtualization exists, those customers don’t want to have to support multiple platforms, they want to support a single system that can deliver both Windows and Linux apps without having to make sure that end users have both clients, that both holes are poked on the firewall, that both gateways are operational, and that both brokers are configured.

The conversation about whether or not there should be Linux desktop virtualization is irrelevant. If customers are asking for it, it should exist. Should you use it? That’s up to you. If you’re just trying to circumvent licensing, then no, probably not. The real way to decide whether or not to use Linux desktop virtualization is to look at your applications. Do you have a need that supports putting them in the datacenter? If so, you should look into it. If not, carry on.

Nobody is making strategic decisions anymore about which OS to deploy company-side. You’re not a “Linux shop” or a “Windows shop.” You make the determination on which OS you use based upon which applications you need. If you have Linux apps, deploy Linux. If you have Windows apps, deploy Windows. If you have both, guess what? Deploy both.

It doesn’t matter if it’s a niche of a niche. If you need it, go get it. There are a number of solutions, and with Citrix and VMware jumping in they’ve brought with them an infusion of energy. As for @AppDetective’s statement that we don’t care about it—it’s just not in our wheelhouse. Clearly there’s a need for it, and there always has been to some degree, but we’re hardly ever asked about it. My opinion, and I think I speak for Brian, too, is that any desktop virtualization is good, and if you have Linux apps and desktops you need to virtualize, we’re all for it. The vast majority of the activity in this space, at least when it comes to operating systems, is around Windows, and that’s where we exist. This stuff is interesting, though, so we’ll keep an eye on it.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Seriously, VMW is going to copy Citrix again?  Wasn't it enough that the VMW keynote copied the Synergy keynote except for a change of 1 word????

Linux VDI is kinda cool though.  


do you think people that want to use Linux for cost saving will accept to pay 300 USD for virtualisation delivery ?

will remain a niche for large entreprise only...


Agreed, Kata, but I think those aren't the people asking for it. I think it's large enterprises that want to deliver Linux apps or desktops in the same way they deliver Windows apps and are tired of using the standard old methods.


I've been using the Linux desktop since 1996 (after using it for 1 year in text-only mode) and I have to say the Linux desktop (which one?  Mostly KDE)  is darn good... and for me anyway, the native application selection covers my usage 99% of the time.  I occasionally need to use a Mac or a Windows box but not too often thankfully.

While there don't seem to be any solid numbers on how many Linux desktops there are, a ballpark figure is in the couple-of-dozens of millions... not necessarily all in the workplace or enterprise.

Red Hat has been doing Linux VDI (and Windows too) for a couple of years now with their RHEV. I have no idea how well RHEV is doing but I'd guess probably ~ 1% marketshare if that.

The SPICE remoting protocol works well enough over a LAN but for non-LAN stuff I've really been enjoying x2go... which is a fork of the GPL'ed NoMachine's NX3 libraries with a bit of new code on top.  x2go has a connection broker component but I haven't tried it and I'm not sure how mature it is... but it continues to be developed, much of it being funded by paying sponsors.

It is worth noting that X11 is quite old these days and was designed before multiple sockets/cores, remoting, compositing, and multi-head... but has been adapted to do most of that stuff fairly well. As a result there have been quite a few core X11 developers who were tired of the ancient design and the quilt-work of stuff on top. So they have been working on a replacement for X11 for several years now named Wayland.  Canonical didn't dig Wayland for some reason so they have been working on a simliar thing named Mir.  It will be another year or two before mainstream Linux distros have adopted Wayland (and maybe a year or two after that for "enterprise" distros) but it is a major change.  Wayland/Mir will have an X11 compatibility layer as long as needed... and Wayland comes with RDP built-in.


I don't think anybody's opposed to VMware doing Linux VDI. The question being asked is: why are they bothering? Given the alternative cost, is this the best use of their resources? But hey, it's their dime ...


Yeah, Dan, I get that. I can only assume that VMware (and Citrix, for that matter) have customers asking for it, so they decided it is worth it.

In VMware's case, Desktone already had support for it. It might not be that big of a deal to make it work elsewhere.

@Scott - thanks for the info on Wayland and Mir. As usual, you've given me more interesting stuff to chase down!


I can't speak so much for the private sector but I know VMware are hounded by the public sector for Linux VDI support.

We use VERDE to provision 200 Linux desktops... I'm looking forward to the day I can use VMware to do this and decommission yet another bespoke system.

I spoke to a lot of crazy Russians at VMWorld - mainly from software development shops who are desperate for Linux VDI support with View.

So I think there is a demand worthy of the R&D and as Gabe just said... Desktone already support it so don't think it's that big of a deal for VMware.

It terms of these throwaway comments about VMware copying Citrix.... Technically with Desktone they had Linux support before Citrix their just making it more of a focus now and at the end of the day so what?


I just wanted to mention that Presentation Server for UNIX is still around. EOL is yet to be announced.  It, along with Web Interface on NetScaler, remains a viable (admittedly niche) solution for Windows-agnostic environments.  As a sidenote, it's unfortunate that the replacement for Web Interface runs only on Windows (There is still no StoreFront on NetScaler).


Uggh! Helmet in lap, thinking more time to wait for things that we really need.


I've got customers who want Linux VDI because WIndows VDI can now do 3D rendering, but a lot of their 3D modeling applications are Linux based. Without being able to do both, doing Windows VDI would result in 3D files in different places for users to manage. So while Linux VDI has been talked about for a long time, and as such some of us are firm in our belief that it doesn't make sense to invest in VDI to deliver a "cheap" OS, the truth is for some customers if we can't do Linux VDI, then they can't do WIndows VDI either. So WIndows VDI opps are being lost because only half the need is being met. That being said, regarding the 3D aspect of a Linux VDI session, I don't think 3D rendering is going to be in the first iteration of support. But it has to start somewhere.


Take a look at BoxedVDI, they have a real nice VDI product, supporting all Linux distro's which are also supported by Virtualbox. There unique IO support makes It possible to run VDI clients without the need to have VDI agent within Client operating system.


Not so much VMware copying Citrix, but Citrix and VMware copying those who already do it, and have been offering VDI for more than 10 years. NoMachine is the leader in this sector, niche or not. Miles ahead of the pack. Let them copy.


"There are two types of potential Linux desktop virtualization proponents: the cheapskates and the people with real business needs. The cheapskates look to Linux because of favorable licensing, and the most ambitious of them can even do it for free. "

Which is why solutions such as x2go are mentioned above, They are absolutely no solution for a firm that has real business needs and uses Linux because it has to, which I might add is becoming more of a norm. My current, and previous employer, ran their entire R&D environments on Linux. It is not uncommon for companies to look at Linux solutions which are free, created by OSS zealots, to then turn to the ones that actually understand companies' real needs. And I am speaking from experience.