IETF 70 Softwire WG Meeting Minutes
Date: 12/3/07
Location: Vancouver
Chairs: Dave Ward, Alain Durand
Scribe: Chris Metz
Meeting Minutes
===============================
 
Meeting opened with usual agenda-bashing and administrivia. 
 
Next, Eric Rosen described lack of progress on interacting with security liason to resolve
softwire mesh security issues. Sam Hartman, Security Area AD explained that a new
security liason would be assigned immediately to work with Eric to close the loop
on this matter.
 
Next, Florian Parent presented an update on draft-ietf-softwire-security-requirements-04. The talk 
included a summery of the rev 03-to-04 diffs and various other sundry remarks. Proposed
next steps were to take this document to the IESG for last call.
 
Next, Yong Cui presented protocol solutions for supporting multicast in a 
softwire mesh scenario as described in draft-xu-softwire-4over6multicast-01. These 
solutions focused on the E-IP-IPv4 and I-IP=IPv6 scenarios for the 1:1 cases where
one IPv4 customer tree is mapped to one IPv6 backbone tree. The draft describes
the use of PIM-SSM and PIM RPF-vector to acheive this. 
 
Upon conclusion of the presentation, Alain Durand asked why is this work being handled in
Softwires. Why not L3VPN? This precipitated a interesting discussion summarized by key 
points made by the following:
 
- Eric. MVPN developed to support many customer (VPN) multicast states aggregated onto one or fewer provider multicast states. The notion
of 1:1 mapping between one customer and one provider multicast state was rejected by L3VPN back in 1998. However Internet Multicast
model and what Softwires is attempting to achieve (that is Global Internet Multicast of one E-IP transited across an I-IP backbone) might
be permissable.
 
- Yakov. MVPN permits 1:1 mapping and supports aggregation of state (many customer trees mapped to one or fewer backbone trees). Thus MVPN
solution could address 1:1 Internet multicast model and as an option provide aggregation of multicast state.
 
- Yong Cui. We have IPv6 backbone that must interconnect global IPv4 sites for unicast and multicast. We have no interest or requirement for
VPNs.
 
More discussion on this topic followed until Dave and Alain suggested that a problem statement for softwire mesh multicast outlining the
problem that needs solving and the requirements that should be addressed be developed and discussed at the next IETF meeting.
Volunteers were solicited and these folks accepted: Yong Cui, Greg Shephard, Yiqun Cai, Chris Metz and Mingwei Xu. 
 
Another brief discussion broke out on the use of global routing handled by L3VPN. Yakov and Eric explained that it was supported
as is today with current L3VPN protocols (e.g. global routes can be placed in VRF), while Mark wondered aloud if that was indeed
the case. It was concluded that if someone did see a need to document/explain how and why global routing should be
handled by L3VPN, that it should be brought to the L3VPN WG. Discussion then ensued on IETF cross-WG procedures,
why Encap SAFI was brought to IDR, Yakov suggesting it could be returned to Softwires, Dave pointing out that
Yakov originally stated it had to be given to IDR and so on. 
 
Next, Alain asked if anyone was interested in working on Phase I of Softwire H & S using L2TPv3. Recall that Phase 0 uses L2TPv2. No one
responded but chairs suggested moving to the mailing list for further discussion.
 
Meeting adjourned.