MBONE Deployment WG (mboned) 
IETF-74                     
MONDAY, March 23, 2009

          
Chairs: 
Marshall Eubanks <tme@multicasttech.com>
Greg Shepherd <gjshep@gmail.com> 
Leonard Giuliano <lenny@juniper.net> 

Scribe: 
Marco Rodrigues 

- Review and status of work items
        
  - AMT draft officially expired today (03/23/09), waiting on UDP draft)

- draft-ietf-mboned-session-announcement-req-01 - H. Asaeda

  - Lenny Giuliano - concern that network based discovery is unnecessary 'today'. There are many tools and content
already out there. SAP/SDP were tools of the past, where today's tools are more than capable. 
  - Greg Shepherd - we need to make sure SAP and SDP are kept separate. SDP is still very applicable, SAP may not be. 
  - H. Asaeda - Main difference between unicast and multicast content, multicast is more serious to create and add scoping
policies and definitions. If there is no scoping configuration or policy then the content has geo scope implications. 
  - Greg Shepherd - I do disagree on the requirements between unicast and multicast and that one is more dynamic than
another is just an assumption. Maybe more fit for another working group such as mmusic.

- draft-ietf-mboned-mtrace-v2 - H. Asaeda

  - Bill Fenner - I support this (removal of response address field). 
  - Lenny Giuliano - Is this documented (reasoning/justification) for removing this? It would be great to note this for
historical reasons for future readers/outsiders.
  - Ron Bonica - Are the responses rate-limited? Unicast, ICMP is limited, but are these rate-limited?
  - Bill Fenner - If the worry is to limit the control plane load, then that's reasonable. 
  - Marshall Eubanks - Can I spoof the responder's address?
  - Bill Fenner - LAN scope has been defined, so you would only be spoofing the address of someone on the LAN. I have no
problem with rate-limiting these messages either. 
  - Ron Bonica - How do you ensure the packet is from the LAN segment? Looking at the link (l2) or the address information
(l3)?
  - Bill Fenner - Wouldn't it be coming in on a different link?
  - Marshall Eubanks - uRPF is not universal by any means. 
  - What not make use of GTSM to ensure packets are only sent from local LAN segment entities.
 
- draft-ietf-mboned-lightweight-igmpv3-mldv2
        
  - Lenny Giuliano -  let's be clear about what the problem is. You have an igmpv3 lite router and you have a igmpv3 full
host, if host sends igmpv3 with exclude, what should the router do?
  - Bill Fenner - by ignoring excludes, do you mean you just ignore the exclude 'content' and not the entire message?
  - Bill Fenner - I wanted to make sure that the actual semantics was that the source-list in an exclude message was
ignored, not the exclude message itself is ignored.  In other words, exclude(S) is treated as exclude(none). A host already has
to be prepared to do the filtering itself, since e.g., another system on the same LAN could have done include(S) or
exclude(none).

- draft-atwood-mcast-user-auth
  - Greg Shepherd - So what does this provide that standard SSM and RPF procedures in a multicast paradigm doesn't already
facilitate? 
  - Lenny Giuliano - revision of IGMP is really difficult to see happening, full adoption of igmpv3 hasn't even happened
from v2. Also, there are knobs certain implementations provide to control access security already.
  - Greg Shepherd - provide clarity on 'security attacks', we need to know what we're trying to solve. 
  - Lenny Giuliano - show of hands to see this item adopted as a WG item (2-2.5 hands for, 0 against adoption)