Multipath TCP (MPTCP)
Meeting  : IETF100, Tuesday Nov 14, 2017, 15:50 - 17:50 (Afternoon session II)
Location : Collyer
Chairs   : Philip Eardley
           Yoshifumi Nishida
AD       : Mirja Kühlewind
URL      : http://tools.ietf.org/wg/mptcp/
Note Taker: Stuart Cheshire

--------------------------------------------------------------------------

1: WG Updates - Chairs

   Yoshifumi: we have a plan to running an adoption call at tcpm.

   ????:
       Why we cannot take it at mptcp?
       
   Mirja Kühlewind:
       The draft-bonaventure-mptcp-converters document is applicable to all TCP options, not only MPTCP.
       The plan is to have discussion in tcpm WG about whether to adopt this.
       If tcpm does not want to adopt this, will bring it back to MPTCP,
       but then it will end up being more narrowly focussed on supporting only MPTCP.
       
   Yoshifumi:
       We will explore the future direction of the draft at tcpm.
       The motivation is to try to get feedback on how to proceed it.

--------------------------------------------------------------------------
	  
2: Considerations for MPTCP operation in 5G - Debashish Purkayastha


   Stuart Cheshire:
        Multiple changing IP addresses on an interface is a good argument
        for having devices support MPTCP, so connections can survive address changes.

   Mark Townsley:
        Multiple addresses per interface is not unique to mobile networks.
        RFC 7934 discusses multiple addresses per interface.

   Debashish:
        in 3GPP, there's a branching point that collects multiple anchors to face mobiles

   Mark Townsley:
        Is it TR23.501?

   Debashish:
        Not sure for now.

   Yoshi:
        Fabien proposed to put community ID in add-addr.
	I will send the pointer of the draft later.

   Yoshi:
        How mptcp layer can know which address belongs to which interface?

   Debashish:
        Implementation of UE or some other parts will provide interfaces


--------------------------------------------------------------------------

3: A proactive approach to avoid performance degradation of MPTCP - Jing Zuo

   Mirja Kühlewind:
        When using BBR, are the congestion controllers coupled?

   Jing Zuo:
        No, not coupled.

   Mirja Kühlewind:
        Why?

   Jing Zuo:
        BBR is not window base.

   Mirja Kühlewind:
        The presentations for BBR at ICCRG might provide useful info.

   Anna Brunstrom:
        Other schedulers have been proposed that avoid the head-of-line blocking
	You might want to try some of them. You might get different results.

   Stuart Cheshire:
        In this presentation you use the word "performance" as synonymous with "throughput".
        Maximum throughput is not the only kind of performance that matters.
        It would be good to use the word "throughput" when that's what the presentation is talking about.

--------------------------------------------------------------------------

4: SOCKS Protocol Version 6 - Vladimir Olteanu (Remote Presentation)

   Yoshi:
        Is token only for idempotency?

   Vladimir:
        Yes.

   Yoshi:
        We don't have to use it when idempotency is not an issue?

   Vladimir:
        Correct. This is entirely optional.