Hi Saku, > I'm thinking if 'VPN forwarder' is catering to many hosts, I might want in > L2 to have top tag for host and bottom tag for guest. If XMPP just allows > single tag, it may exclude external VPN forwarder or at least limit its > scale.
Oh I see ... it could be a valid case. But I am not sure why in this case we could not just have flat single VLAN tagging for guests between hosts and external VPN forwarder. Would you expect more then 4K (VLAN limit) to be exceeded ? If not I guess there is no good reason for QinQ. In fact if you have QinQ there the semantics of VPN label used by such VPN forwarder would need to include direct demux for both VLAN tags wouldn't it ? Cheers, R. On Tue, Apr 2, 2013 at 1:48 PM, Saku Ytti <[email protected]> wrote: > On (2013-04-02 12:42 +0200), Robert Raszuk wrote: > > Hi Robert, > > > Do you have in mind sort of CSC analogy that tenants internally have > their > > own demux based on bottom 802.1Q tag while top one would be used just > like > > a single VLAN tag as described ? > > I'm thinking if 'VPN forwarder' is catering to many hosts, I might want in > L2 to have top tag for host and bottom tag for guest. If XMPP just allows > single tag, it may exclude external VPN forwarder or at least limit its > scale. > But maybe that is intentional, to steer away from external VPN forwarder in > large-scale deployments. > > > I think you can build IPv6 DC virtualization services while still using > > IPv4 only DC core. > > I was just worried as draft specifically allows for IPv6, yet XMPP messages > seem to be trailed with /32. I'm personally not interested much in > dual-stack control-plane. I'm happy to run 6PE now, and maybe in future go > IPv6 only control-plane and 4PE for legacy. But I understand it might not > be popular position. > > > The translation happens at the option B ASBR. I would not really run MPLS > > LDP + IGP or IGP with extension to carry labels or other form of tags in > > the DC. > > Fully agreed. I could see myself running this draft, but I would not want > to trust 'VPN forwarder' fully, so OptB with appropriate implementation > which actually does implement label checking would be requirement for me. > > > Personally I like Petr's draft: > > http://tools.ietf.org/html/draft-lapukhov-bgp-routing-large-dc-03 > > I'll have to read it, thanks. > > -- > ++ytti >
