Regarding the big infrastructure software vendors [1][2] 2010 will be the year of VDI. The key to success will be the client-hv…
OK, but what OS will be used on the virtualized clients? Will it be Windows7 only or will it be a mixture of Windows7, MacOS and Linux? Or will it be a mixture of Windows7, MacOS and ChromeOS? One could say: Whatever you need
But wasn’t the idea to simplify desktop management what was originally driving VDI…? Admins will change application hazzle against os hazzle – and if you think of laptops with special mobile hardware (umts cards, cams, docking stations, etc.) they will get the os on different hardware hazzle for dessert ![]()
(I’m writing this although Brian Madden already explained that the client hv will be the new hardware abstraction layer [3]. It’s not only about HALs! It is also about different drivers, that have to be installed on top of the os(es) – The client hv can’t do anything about that!)
So on the one hand the IT guys have to get along with the os on different client hardware hazzle and on the other hand they have to maintain the different application delivery infrastructures, like XenApp farms, webservers and ordinary software deployment tools – It’s not getting easier, it is getting more complex!
—
PS: For testing purposes I’ve installed Ubuntu Linux + Sun Virtualbox + Windows XP on my laptop. It is great! But I’m a computer engineer with lots of experience in server and application virtualization – An average user couldn’t handle two taskbars from different operating systems at the same time (see screenshot)…
[1] http://mylearn.vmware.com/courseware/51774/VMworld2009_General_Session_Herrod.pdf
[2] http://www.citrix.com/English/ps2/products/subfeature.asp?contentID=1858899
[3] http://www.brianmadden.com/blogs/brianmadden/archive/2009/03/09/why-client-hypervisors-will-be-a-big-deal-hint-it-s-not-about-running-multiple-vms.aspx


I agree with your view that a client hypervisor alone will not simplify management. However, it is possible to use a client hypervisor as an enabling technology to remove time and complexity from remote PC management. At Virtual Computer, we have build a product that includes both a client hypervisor and a VM management stack. In addition to abstracting the base hardware, we handle all hardware drivers at the hypervisor level, presenting a generic set of virtual hardware to Windows. So, the same Windows VM that is running on a Dell with an Nvidia graphics card now could be running on an HP with an Intel graphic card 15 min. from now with no driver changes required in Windows. That’s also just the tip if the iceberg in terms of management features, so if you are interested to take a look at some point, I would welcome the opportunity to provide a web demo.
I guess that you actually mean that the IT department will have to deal with the drivers for the Client HV layer (I am assuming we have a type 1 client HV, not a type 2 as you are currently using with a Linux host OS and VirtualBox type 2 HV).
For the guest OSes that can run in the guest VMs, no need to care about a lot of different drivers; as these OSs will normally use all the same (virtual) HW if you use the same Client HV on all the physical clients.
And that’s true, the type 1 client HV will run on different HW (even if you use a Mac, they have several HW, which require different drivers).
What I don’t understand in your article is why you think that client HV is important to Virtual Desktops Infrastructures ?
AFAIU, the admins of the VDI DCs do not care about the client devices, as long as they have the connectivity and right “remote desktop” layer, which can even be fired using a Web Browser.
Can you elaborate on the topic of the interaction between VDI and client HV and why the later is a key to the success of the former?
@Mr V: The Client HV will build the bridge to the offline world. In an ordinary VDI you can not take your OS-Image with all your applications with you (i.e. on a business trip) – With a client HV (installed on your laptop) you can checkout your VM, go offline and checkin after going online again.
Thanks for the answer.
It’s clearer now, even if I feel it’s a little painful to check out the VM (so that it can run locally) before taking it with you.
How long will it take to do that?
It depends on your network speed and the size of the VM how long it takes to initially checkout. After the first (full) checkout checkin and checkout should only transfer the delta, so that the process will speedup a lot.
It isn’t a solution for all Desktops, only for hardcore users like developers, administrators, scientists and perhaps some chairmen
@Sebastian, @Doug
How are the Windows licenses used in that scenario? Do you have to use VECD licenses only (expensive and recurrent cost)?
Can you re-use the OEM license you got with your laptop and transfer it to the VM that runs locally or even in the datacenter, assuming that you formatted your HDD and installed a type I hypervisor? In that case you’ll never use your OEM license, you would run only one instance of Windows XP/Vista/7 simultaneously, so it could make sense but will MS allow that?
I guess you can’t use OEM licenses because you are using your image on different machines…
But I don’t know the right answer of this question.