[H]omer wrote:
> Verily I say unto thee, that Roy Schestowitz spake thusly:
>> Will Microsoft Buy the New Citrix?
>
> They don't innovate, they buy ... remember?
>
> Anyway, I wouldn't worry about it. KVM is a better technology (as usual,
> Microsoft chases the weak solution).
>
>> KVM is a better technology <<
I know this can be argued either way, so in the end is only personal
opinion. But still, it isn't easy to come up with a sweeping statement that
declares one better than the other, yet.
KVM could be said to be better because, provided you have the VT chip, it is
a full virtualisation. Truely independant OS's on the same machine,
Otherwise (without the VT chip) it falls back to being paravirtual (as Xen
is).
In Xen though you have a level of shared resources that your hosting systems
can tap into, currently only at the privelidged level, but possible to
extend this. While at the same time the VM host can keep an eye on the
hosted OS, it's only a process like any other after all, so auto recovery
processes are more likely to become available. Then there is talk of
putting drivers and such into the VM host layer.
I know people argue both ways about this too, but why shouldn't there be
common drivers between OS's. If you have a VM of say openSuse and another
for Debian on the same hardware, why not have common drivers, they are
using the same hardware.
Yes, I know, they are less independant which isn't the goal with VM that is
usually talked about, but what I'm saying is that the goal isn't
necessarily full independance. I bet your left sock that once people (other
than web hosts) have got their KVM's, the first thing their have to get on
with is finding ways to marry them together and feeling that it seems silly
using the usual NFS's and NIS's and tunnels etc, when they are all on the
same physical hardware.
|
|
