OSX Virutalization

Bill Bogstad bogstad-e+AXbWqSrlAAvxtiuMwx3w at public.gmane.org
Thu Jan 7 12:22:20 EST 2010


On Thu, Jan 7, 2010 at 11:42 AM, Dan Ritter <dsr-mzpnVDyJpH4k7aNtvndDlA at public.gmane.org> wrote:
> On Thu, Jan 07, 2010 at 11:26:07AM -0500, Jerry Feldman wrote:
>> It is certainly possible to virtualize any OS.
>
> Oh, you'd certainly think so. But when I tried to move SuSE
> Enterprise 8 to a VM (tried KVM, VirtualBox, VMWare) I
> discovered it would crash within a few minutes of any
> significant usage.
>
> I don't think SuSE was intending this behavior, but it was real.

I'm sure it was.  However, you are making a statement about a specific
set of virtualization products
whereas I think Jerry is talking about the theoretical case.   VM
products have bugs just like everything else out there.  I suspect
that sometimes these 'bugs' (imperfect hardware emulation) are left in
for performance reasons.  Emulating hardware devices with 100%
accuracy is likely to be very computationally expensive and if most
OSes don't need some obscure feature/(virtual) timing characteristics,
why bother if it slows things down by 10%.  It may even be that the VM
product matches the hardware documentation exactly, but the hardware
doesn't behave the way the documentation says it does.   OSes written
to the hardware documentation will then work in a 'perfect'
virtualization product
whereas OSes that were written by trying things out on actual hardware
will not work in the 'perfect' VM.

Bill Bogstad





More information about the Discuss mailing list