On Mon, Apr 12, 2010 at 02:01:51PM +0300, Kenneth Nagin wrote: Normally migration is co-ordinated by the libvirt client in a 3 step process Client ------> Dest host (prepare) Client ------> Source host (perform) Client ------> Dest host (finish) THis changes it so that libvirtd on the source talks directly to libvirtd on the dest Client ------> Source host (perform + p2p) Source host ------> Dest host (prepare) Source host ------> Dest host (finish) This means the client does not need to authenticate with the dest host libvirtd. This is for cases where the underlying virt can do migration without any support from libvirtd at all. The hypervisors talk directly to each other Builds on top of p2p, but instead of having QEMU connect to QEMU, the migration data is tunnelled over the existing libvirtd -> libvirtd data coonnection which is more secure. The config file for the guest is written out to disk on the destination host making the guest persistent. Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|