I've been running Type 2 hypervisor apps for years, and whether it's VMware Workstation, Virtual PC, or Parallels, each has proved vital in allowing me to test and run a wider array of applications and OSes off my client machines. But what of the notion of running a bare-metal hypervisor on the client? What benefits might that bring?
As it turns out, quite a few -- some of which could point the way toward a new future in client-side computing.
The most obvious upside of bare-metal hypervisors is that they abstract the client OS from the underlying hardware. In the process, these so-called Type 1 hypervisors would allow a single Windows disk image to run across a variety of devices without worrying about the HAL (hardware abstraction layer) that typically separates the OS from the bare metal.
Moreover, when combined with a solid methodology for pushing virtual images down to clients easily, bare-metal client hypervisors could eliminate the stress of maintaining a thin-client environment, where continuous connectivity to the server is required.
Security would also be greatly enhanced. Imagine two OS instances running side by side -- one a locked-down enterprise version and the other configured for personal use -- and you can quickly see how handy bare-metal hypervisors could be on corporate laptops. Factor in encryption, and the virtual system could be rendered inaccessible (sort of like a VM BitLocker) in the event that the laptop is lost. Or, as is the case with today's mobile devices, it could be remote-wiped automatically when reported stolen.
As might expect, a number of companies are at work in this promising space, including Neocleus, Citrix (in conjunction with Intel as part of Project Independence), Xen, and VMware (also in collaboraiton with Intel). But it was relative newcomer Virtual Computer, with its NxTop, that opened my eyes to the powerful promise of bare-metal client hypervisors.