On 2007-09-08, 14:59 GMT, Fernando Cassia wrote:
> The above move by RedHat is a bit confusing... what can Xen do 
> that KVM cannot?. In other words, why should anyone even bother 
> with Xen with KVM around ??.  I've read Xen is "more robust" 
> because it has a "one year lead" over KVM. But really, how does 
> this translate, if performance of Xen could be worse due to 
> more paravirtualization?.  Or is Xen more optimized to
> provide the "greatest possible consolidation" on servers (ie 
> less resource usage, less impact on cpu usage of a high number 
> of VMs?).

(notwithstanding the domain of my email address, I am not part of 
the kernel/VM RH team,  I am not developer or even programmer, 
and this is just my personal thinking not anything close to 
official position of Red Hat).

a) Concerning maturity -- remember, RHEL 5 was created on the 
   basis of Fedora Core 6, which was released in the fall 2006.
   In that moment, there was no KVM around. And you don't put 
   technology which is not enough old into product which you are
   supposed to support for the next seven years. It is almost one
   year later and KVM is still in the stage of rapid development
   and substantial changes (IMHO and I think it is a good for all 
   of us -- eagerly waiting for KVM host suspend/resume in
   2.6.23).  Again, you don't put such stuff into enteprise grade
   distribution.
b) Concerning paravirtualization, I think you have it other way
   around -- given the fact that both host and guest know about
   virtualization, they can modify their behavior accordingly and
   work better than with full virtualization, when guest knows
   nothing. It is said that the disadvantage of fullvirt is
   slightly smaller with hardware accelerated virtualization
   (which is what KVM does), but again I know nothing about this.
c) One thing which is very important to consider is libvirt.  
   I think that is really smart move from Red Hat. Remember,
   all interesting tools for doing virtualization (virt-manager,
   virsh, virt-viewer, etc.) in Red Hat distributions are based
   on libvirt and so they agnostic vis-a-vis particular
   implementation of virtualization. So our users can learn about
   virtualization as such with Xen, and when some other
   technology (e.g., kvm, but who knows what will happen in seven
   years and whether some even better technology won't arise?) is
   good enough to be included into RHEL (for example, that
   actually missing paravirtualization is one of the biggest
   problems of kvm) or if something nasty happens to Xen, we can 
   just switch the virtualization backend and everything will
   work for our users as before.  Again, I think this is the way 
   how to do a seven-years-supported distribution.

Just my 0.02 CZK.

Matěj

-- 
http://www.ceplovi.cz/matej/blog/, Jabber: ceplma<at>jabber.cz
GPG Finger: 89EF 4BC6 288A BF43 1BAB  25C3 E09F EF25 D964 84AC
 
A bird in the hand makes it awfully hard to blow your nose.


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Reply via email to