Richard W.M. Jones wrote:

Hugh Brock wrote:

...

Daniel P. Berrange wrote:

...

3. The way I think you re suggesting - a libvirt server on every remote host which calls into the regular libvirt internal driver model to proxy remote calls. So even if the hypervisor in question provides a remote network management API, we will always use the local API and do *all* remote networking via the libvirt server

http://people.redhat.com/berrange/libvirt/libvirt-arch-remote-2.png

This strikes me as *much* easier to manage, and the most consistent thus far with the idea that libvirt should remain as hypervisor-neutral as possible.

I guess the management issue is going to be versioning the protocol. If the protocol is just a direct mapping of vir* calls and structures then you'll quickly end up in a situation where even the smallest change requires you to upgrade the world or old versions have to be maintained indefinitely.

That's not saying I don't like the idea.

True enough... we're guaranteeing we're going to have backwards compatibility problems. On the other hand the libvirt API is supposed to be held pretty stable. DV, any thoughts? --H