Tom, > >> Nothing against leveraging BGP. Doing BGP at end-hosts is clearly a > >> problem. XMPP is an > >> option, but not the only one. > >Some providers are looking for a solution where network virtualization > >forwarding and control functions are fully decoupled, and where only > the > >forwarding function (i.e., overlay encapsulation function) is > implemented > >on, e.g., the application servers. > > I am confused. Why do you not think that control and data plane > are not > separated in this solution?
Because the draft proposes that NVE implements both the control plane and data plane functions of E-VPN PE. Maria _______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
