On Tue, Sep 09, 2008 at 01:03:17PM -0600, Stephen John Smoogen wrote:
On Tue, Sep 9, 2008 at 12:40 PM, Fabian Arrotin fabian.arrotin@arrfab.net wrote:
Karanbir Singh wrote:
admin wrote:
So when can we expect KVM/Ovirt to ship as RHEL/CentOS's default virtualisation solution?
oVirt itself is in beta now - you should be able to get pkgs and try it out NOW. and for most part, it works fine ( in the small scope of testing that I've done personally )
- KB
Now i understand more why Red Hat is touring (in Europe at least) and busy promoting Virtualization : it seems (if i believe the agenda : http://www.europe.redhat.com/promo/business-partner-training/agenda.php) that oVirt is becoming the tool they want to promote (even more than libvirt in a standalone fashion) , at least that was my perception (especially when you read the event invitation email) .. let's see what the future will be but all people following the Fedora line already saw that kvm became prefered over Xen ... and from a Market-Share point-of-view it's also clear that RH had to push something new to compete against Xen (through Citrix XenServer), Vmware, and HyperV ...
Actually Red Hat didn't need to do that for competition reasons. The big problem is that Xen is stuck on an old kernel and not in mainline. It was getting harder and harder to forward port the kernel mods and after 2.6.26 basically became impossible. If Xen spent more energy working inside the main kernel.. I don't think the KVM strategy would have been followed.
Xensource seems to have a couple of developers working with upstream Linux now.. upcoming Linux 2.6.27 will have more Xen domU pv_ops features (64b support etc), and they're working with getting Xen dom0 pv_ops support merged in.. some groundwork already sent/submitted for 2.6.28.
More information about upstream/pv_ops Linux Xen kernel: http://wiki.xensource.com/xenwiki/XenParavirtOps
Too bad they didn't start working with that stuff earlier :(
-- Pasi