Why users don’t want a client hypervisor (warning, opinion inside)

At Citrix Synergy last week Citrix talked up their client hypervisor and even released XenClient for you to test.  Rumour has it that the next version of VMware View missed a ship date to wait for VMware’s client hypervisor (I have no inside knowledge and am making up rumours as I go).  Of course Virtual Computer’s nxTop has been a shipping client hypervisor for months. Client virtualisation is the new desktop virtualisation.  It’s being talked of as a great class of product with a big future, but it’s done wrong.

A lot of people seem to think that a client Hypervisor enables employee owned IT, or Bring Your Own Computer (BYOC).   Employees are given an allowance to buy the PC they want to use and IT will deliver the desktop they need for work onto that computer.  These people miss the critical part.  When I’m a user who brings my own computer to use at work I don’t want to have to re-install my own OS.  I don’t want IT to remove my installed OS and give it back to me in a VM.   I want the OS that my PC vendor designed to work with my shiny new toy, I want all the bells and the whistles, even the flashing lights are important. This means that I don’t want a bare metal Hypervisor with limited hardware and feature support.

I want IT to provide me a package that is their desktop, this means I want a hosted hypervisor.  I want that package to have a light footprint on my computer and to be easily removable.  I want my OS of choice to own the hardware and the corporate machine to live inside my OS.  Personally I’ve been looking for a way to put that package on a USB key I can use anywhere for three years and am still a long way away.

The client hypervisor in it’s current form is for an IT owned asset that runs an IT owned desktop.  It might also let me run my own OS but that’s a secondary use case.  The client hypervisor is installed before any OS, mine or IT’s.  Next IT installs the corporate OS.  If I’m very lucky I might be able to run my own OS.  If I’m super lucky I might be able to run my OS at the same time as the corporate OS (XenClient can, but IT might not let me).

So when will users want a client hypervisor?  When they don’t need to know it’s there.  If the hypervisor is an invisible layer under the OS they bought and IT can use a standardised way to deploy their corporate VM to the employee’s PC without affecting the installed OS.  This hypervisor needs to be as invisible as the BIOS on their laptop and it probably needs to be more hardware specific than the BIOS, it needs to know the Video card and the sound card so it can virtualise these for my OS and for the corporate OS.

The problem I see is that there needs to be a standardised, vendor neutral, way to build a corporate VM that runs on this hypervisor but the actual hypervisor needs to be hardware specific, but which hypervisor to choose?  If Apple make their own Xen KVM and Dell choose VMware and HP choose XenClient what then?

Call me cynical (or old fashioned) but I don’t see a future where client VMs can run unmodified on multiple client hypervisors.  I don’t see a future where hardware vendors will have builds of multiple client hypervisors for each of their hardware platforms.  So I don’t see a future where client hypervisors work.

Again in the future I see hosted hypervisors running corporate VMs inside employee owned operating systems on employee owned hardware.  I wish it weren’t so, but I can’t see commercial reality leading anywhere else.

Keep in mind that a key lesson from the success of iPhone and iPad is that IT must work for consumers (idiots with credit cards), not for IT (us Geeks who know better).

© 2010, Alastair. All rights reserved.

About Alastair

I am a professional geek, working in IT Infrastructure. Mostly I help to communicate and educate around the use of current technology and the direction of future technologies.
This entry was posted in General. Bookmark the permalink.

6 Responses to Why users don’t want a client hypervisor (warning, opinion inside)

  1. Great point of view. You articulately define the two use cases:

    1. Employee BYOC then use hosted hypervisor (ACE?)
    2. Employer provided device then possibly use client hypervisor

    This just adds to the morass of complexity that is VDI and further confirms what us non-VDI Kool Aid drinkers though all along: the last people to be consulted are the End Users. Assessing the End User is not the same as Consulting the End User.

    You see this in IT where “everyone but the IT guy” uses VDI. I’ve heard senior execs at large customers describe VDI as a “career limiting move”.

    Another fine post in the VDI consciousness and I recommend all VDI aficionados to have a cold shower in this.

    Great work,

  2. Craig Miller says:

    I think your not looking at this the right way. I agree with you that that if I follow the BYOPC model, giving the IT dept my computer and getting my old OS wiped away with limited support, but this isn’t the only way inwhich a client hypervizor can be delivered. I saw a demo of the Neocleus product at Briforum ’09 which showed a way of deploying an MSI to systems, which would then bury itself under the OS on install. It would then do a P2V of that system and do direct hardware transparency to the OS so in the end for the user you get the same system you had before, with all of the same hardware that was supported before. The inky difference is that now the IT dept could now deploy another OS to run in parallel on my system. Tell me the shortfalls of that!

  3. Alastair says:

    I like that Neocleus deployment model, I’d be very interested in seeing the end user experience, OS and hadware support.

  4. Alastair says:

    “Morass of complexity” = “new technology where complexity is not yet hidden”

    I like VDI as a solution to situations where datacentre based end user computing is required and Remote Desktop Services isn’t acceptable.
    I use a VDI desktop so I don’t need to take my laptop to the training centre where I’m teaching, just my USB key with the View Client on it.

  5. Citrix was actually pretty careful to distinguish between these two use cases at the launch of XenClient. Their vision of the BYOC use case doesn’t use XenClient at all. They see you installing Receiver on your laptop (or MacBook or phone – that’s why they call it BYOC and not BYOPC) and talking to a XenApp or XenDesktop back-end. Desktops run remote, or applications run inside a capsule. (Their Mac integration for XenApp Windows apps is not bad.)

    XenClient, in this view, is for corporate-owned hardware, where IT can provision both a personal and a business workspace – and a dev or locked-down one as well. This was the model which they demo’ed on stage, and which Simon Crosby has been writing about on his blog.

    Neocleus does demo well but they’ve pulled back a bit from enterprise sales and are now hoping to OEM the platform to partners like BigFix. Virtual Computer is still trying to push NxTop into the enterprise market.

  6. Capriole says:

    I think the comments are accurate for the use case you describe, but I think there is an important alternate use case about making corporate support of the desktop more efficient.

    Still IT owned hardware, IT managed hypervisor, and IT managed OS.
    But now IT can move that OS to new hardware without changing the user experience.
    They can also provide multiple OS versions on a single piece of hardware – great for simplfiying OS migrations.
    Backup is easier because we can snap the entire VM when it’s on-network.
    The OS build becomse simpler (only virtual hardware to support) although admittedly that complexity just moves to the hypervisor.

    None of this has any direct impact to the end user (and only minimal tangental benefits to them) but it does have huge benefit to IT shops managing large desktop fleets….

Comments are closed.