Red Hat (finally) releases desktop virt product: Red Hat Enterprise Virtualization for Desktops

We got to take a look at Red Hat Enterprise Virtualization for Desktops at BriForum a few weeks ago, but they saved the release of the product for their own Red Hat Summit event last week. As you probably know, this is the long anticipated re-release of the SolidICE product that Red Hat acquired from Qumranet in 2008.

We got to take a look at Red Hat Enterprise Virtualization for Desktops at BriForum a few weeks ago, but they saved the release of the product for their own Red Hat Summit event last week. As you probably know, this is the long anticipated re-release of the SolidICE product that Red Hat acquired from Qumranet in 2008. RHEV for Desktops uses the KVM hypervisor that is built in to the Linux kernel combined with their SPICE protocol to deliver a high quality user experience.

Brian and I did a side-by-side comparison before the acquisition, and we learned that SPICE had some advantages over the competition, along with some drawbacks, like being tied to a specific hypervisor (see Brian's article on remote protocol hypervisor integration). Now that Red Hat has had over 18 months to enhance the product, we'll be getting it up in the lab soon to see what's changed. Keep your eye out over the next month or so for a look at the management and protocol capabilities.

Reader Scott Dowdle (who I consider to be our open-source watchdog, making sure nothing on that front gets by us :) sent in a few details related to the desktop product that we might have overlooked:

1) The RHEV-Management app now runs on Microsoft Windows 2008 Server R2 rather than 2003 as 2.1 did

2) They have added a V2V conversion tool that can convert from ESX, RHEL Xen, and RHEL KVM to RHEL KVM and RHEV

3) The underlying technology for V2V is libguestfs, which is a library that has bindings for many different programming languages to access VM disk images and interact with them in a variety of ways. Some of them work while the VM is running, and others work while the VM is powered down.  Various command line apps are available that use libguestfs including guestfish, and many virt-{something} commands. 

4) They showed an early version of a GUI application that uses libguestfs

(Chetan Venkatesh added a comment below that describes how you can use libguestfs)

Licensing boils down to two different licenses (which are really maintenance packages) that must be purchased. First, you're required to have RHEV for Servers licenses on each host system. Then, you'll have to purchase the appropriate number of RHEV for Desktops 25-user license packs. Desktop licenses are concurrent, though, so many companies will be happy with that.

Two "starter packages" are available, and each includes six sockets of RHEV for Servers and 100 RHEV for Desktops licenses. The difference in pricing is based on what support level you need. 12x5 support runs US$4,494, and 24x7 support runs US$6,744. Since these are actually maintenance contracts, this cost is per year.

Based on the costs of RHEV for Servers (here), which are $499/yr for 12x5 support and $749/yr for 24x7 support, we can get a good idea of what the 25-user packs will cost. Using 12x5 support as an example, we see that $2994 of that is for RHEV for Servers ($499 x 6), which leaves $1500 for 100 desktops, or $15/desktop/yr. The same math on the 24x7 package results in a cost of $22.50/desktop/year. So, a 25-user license pack would cost either $375 or $562, depending on the level of support.

Since we're not fans of generalized VDI cost models, it's not really worth the time to plug these numbers into any to see how it comes out. I'm curious to see what these costs would amount to if they were placed into an actual environment, though. So, if you'd like to weigh in on how this would affect your organization, please post your thoughts and findings in the comments.

We're still trying to put together the plans for our next Geek Week, and a protocol shootout between HDX, RDP (with and without RemoteFX), SPICE, and EOP is one of the options. If you'd like to see that (or something else, for that matter), let us know.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

libguestfs and guestfish are awesome tools and in the right hands more powerful than anything I've seen possible on powershell/powercli.

For example - If I want to see the size of the system32 directory of a windows Vm which is not powered on, I can simply -

$ echo du 'win:C:\windows\system32' | guestfish --ro -i Win2003x32


I can copy a file into a VM without needing to power it on, set its networking etc etc

$guestfish cp

the coolest one I use is is the guestfish support for inotify handles that will trigger if a certain file I want to watch for in a VM changes - all this without having to *** around with in guest tools (vmware tools)VMCI/code/etc

For those not averse to the linux cli  - playing with KVM, libvirt, libguest and guestfish can teach one a lot about virtualization and operating systems design.

Gabe thanks for actually talking about the guts of the Redhat offering.

Chetan Venkatesh

CTO & Founder

Atlantis Computing


Wow, that's pretty cool! Thanks for the explanation. Credit for bringing it up goes to Scott, though, because he's the guy that pointed it out to me.

I'm leaving for vacation in a week, but I'll lab this up when I get home.


I'm really looking forward for a video review/hands-on/no-bull demo on RHEV RTM /w some RH tech and Gabe/Brian. Hope it wll happen sometime soon. Ya know, I really love those geek video things you guys make.

As a side note: How do you guys see RH as compared to the competition? Is there any validity to the claim that this is a milestone in the rise of KVM vs. XEN in the OSS?


@kimmo IMHO KVM is still a generation or two behind xen in not just enterprise readiness but some basic capabilities.  Xen did not stand a chance without Citrix but with citrix's stewardship has done a lot of good things for  xen.  KVM still has to deal with improving basic io on virtual block devices (the loop file driver implementation on Linux is very slow and inefficient), couple that with the fact that linux does not support thin provisioning in the true sense on lvm2 means that new extensions to lvm2 and device mapper need to be built.  Xen has a lot of good infrastructure like the open switch and fault tolerance via project Remus available which are xen specific, KVM will need these before it can talk about virtualizing any tier2 or tier1 w/loads in the  data center.

I think KVM can get there quickly (18 months) given red hat is exceptional at taking Linux oriented infrastructure to market but neither vmware  microsoft or Citrix xenserver are going to sit around waiting for them to catch up.


@Chetan Thanks for a great feedback on the topic. Much appreciated



What's all this about the loop file driver?

That's not been used by KVM (or Xen or QEMU) for at least a couple of years.

I think you're operating under very old information about KVM's I/O model, but given your companies focus maybe that's not exactly by mistake?

What many folks seem to conveniently forget is that Xen and KVM has very similar base architecture. Xen also uses QEMU as it's backend for I/O - and like KVM does NOT use loopback file driver it uses TAP:AIO by default.

You also say that OpenVSwitch is Xen specific - it certainly isn't. Take at look at where you can get the facts - (xen, KVM and Virtualbox support)

As for Remus - it's a hobby project at best and I've yet to see any organization deploy that tinker-toy.

The only credible F/T solution I've seen so far (and that includes VMware F/T) is Marathon technologies offerings.

I think it's time to have another look at KVM before you spread this fud.


Red Hat Linux, assembled by the company Red Hat, was a popular Linux based operating system until its discontinuation in 2004.

Red Hat Linux 1.0 was released on November 3, 1994. It was originally called "Red Hat Commercial Linux" It was the first Linux distribution to use the RPM Package Manager as its packaging format, and over time has served as the starting point for several other distributions, such as Mandriva Linux and Yellow Dog Linux.

Since 2003, Red Hat has discontinued the Red Hat Linux line in favor of Red Hat Enterprise Linux (RHEL) for enterprise environments. Fedora, developed by the community-supported Fedora Project and sponsored by Red Hat, is the free version best suited for home use. Red Hat Linux 9, the final release, hit its official end-of-life on 2004-04-30, although updates were published for it through 2006 by the Fedora Legacy project until that shut down in early 2007.