V6ops meeting notes - Fred Templin
IETF-55 Atlanta
19 Nov 2002

***************************
1) Unmanaged networks (Christian Huitema)
   - privacy analysis - privacy addressing issues for various environment; BCP document needed
   - better NAT-PT needed
      - reserved prefix in IPv6?
      - mapping of IPv4 to IPv6 by V6-only host
      - v6-only to v4 netwoks?
   - Comments:
      - keep things simple so that operators can deploy easily
      - Teredo too complex
      - micro-optimizaions needed
      - too complex for corporate LAN mgr.
      - can tunnel broker be used to cross the NAT?

2) ISP - Cleve Mickles:
   - multi-homng; address management
   - overlap with enterprise/managed space?
   - new name: home networking to broadband ether
   - public wireless LAN
   - infrastructure svcs
   - Itojun comment:
     - multihoming is issue for enterprise
     - assigned addresses?

 3) Enterprise/managed - Yanick Pouffary:
   - new mailing list
   - solutions will be part of a seperate document
   - network connected to an Internet provider?
   - Comment:
     - draft needs more text to define scope
     - business requirements may drive IPv6 decision
   - S/W transition points
    -  DNS routing
    - address plan
    - network mgmt
    - IPv6 address scoping
    - Comments:
      - one enterprise (at least) has deplloyed v6 - real deployment carries more weight
      - doesn't like he idea of using vendors' input (wants input from someone like GM)
      - doesn't like the term IPv6 NAT

  5) V6ops-3GPP - Jonne S.
     - V6ops 3GPP design team
     - scenarios doc WG item
     - seems stable
     - analysis doc - editorial changes
       - static vs. dynamic tunneling
       - NAT-PT vs. NAT-64
     - Scenarios:
      - dual-stack IMS scenario
      - Jonne: IPv6-only IMSs
     - Analysis: need to mention dual-stack CSCF in 4.2
     - WG last call for scenarios
     - WG draft for analysis - accept as WG item?
     - NAT-PT issues:
       - NAT-PT needed for IPv6-only nodes
       - Should only be used in stub networks?
       - numerous comments on whether a NAT-PT solution can be made
         to work at all. This is a matter of possible concern for the
         IPv6-only terminal

  6) RFC 2893(bis)
     - dynamic tunnel interface MTU
     - 1380 bytes proposed for MTU when tunnel not dynamically config'd
       (DF bit not set)
     - (1380 = VPN MTU - 20 for IP)
     - reassembly buffers NOT 64K; 4400 is Erik's pick for now
     - ingress filtering?

  7) 6to4 Security Considerations (Pekka Savola)
     - spec is very terse
     - automatic tunneling mechanisms used in same box
     - relay spoofing; anyone can spoof 2001::/16 addr's
       pretending to come from relay
     - relays use RFC 3068 as their source address

  8) Harald Alvestrand
     - v6ops group closing - what to do with NGTRANS drafts?
     - ask for experimental status?
     - go to ADs and ask for stds track?
     - aspects of transition were mis-managed
     - "circuit switching in an alternate reality"?