Notes on VirtualBox

[richard.pieri-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org (Richard Pieri)]

On Apr 18, 2010, at 2:26 PM, Rich Braun wrote:
> 
> * Time-sync with the host works; under VMware the guest O/S clock advances at
> least 10% faster than real-time

In VMWare's defense, this is *not* their fault.  The fault lies in the varying ways that operating systems tell time, none of which are especially reliable on Intel hardware,

http://www.vmware.com/pdf/vmware_timekeeping.pdf

compounded by the inherently flaky nature of x86 virtualization which I addressed previously.


> * There is a relatively straightforward way to export VM images and copy them
> to another machine; VMware Server allows you to do this but there are some
> hoops to jump through, and there's no guarantee you'll be able to import them
> in a future or alternative version of any of VMware Inc's products

FWIW, VMware has been *much* better at this then Parallels.  All of VMware's products of a given generation use the same vmkernel, and that makes horizontal moves quite painless.  VirtualBox is likewise good about this, although you need to do more than just copy the disk images if you don't already have VMs configured at the destination.  Parallels is a total loser.


> * Integration with the openSuSE host O/S is seamless out-of-the-box; with
> VMware server you have to download kernel modules which may or may not work
> with the latest/greatest version of the openSuSE distro, and which prevent you from doing upgrades of the host O/S as they become available.

Last I knew, both VMware and VirtualBox use the same method: compile kernel extensions against the currently running kernel sources if an exact match set of modules is not found.

--Rich P.