On 04/12/2011 08:13 AM, jbarkley wrote:
From: jbarkely<jbarkley@willo.(none)>
James and Mathias:
Have you taken a look yet at my proposal to expand the usage of the
network XML?
https://www.redhat.com/archives/libvir-list/2011-April/msg00591.html
I'm wondering if you hit any limitations while trying to make ESX's view
of networks match libvirt's view, and if I could do anything in the code
I'll be writing to make the two fit together better (note, for example,
that in the new model, you can specify a normal host bridge device, to
be used for connecting a guest "directly" to the physical network, and
likely we'll add other types, such as the 'openvswitch' type I suggested
in (so far the only) followup).
My own last experience with the VMWare network model was when using
VMWare Workstation over 2 years ago, but I recall that for them bridge
vs. NAT vs. routed was a 3 way switch at one level, rather than having
bridging setup off in the Balkans, as is currently the case with
libvirt+KVM/Xen ;-) Also, danpb mentioned yesterday that this expansion
of libvirt's network model may bring it closer to VMWare's.
I will be adding in this code "very soon" (within the next month) so it
may be reasonable for you to put yours on top of mine, rather than
putting in something now, then re-working it later (which could cause
pain if we had to forever support some "interim" config model that we
later decided to change).
Please have a look at my mail and rip it apart. Let me know if there's
anything in there that won't work for you, and if there may be something
else you'd like added.
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list