Re: [5gangip] Side meeting notes

[Behcet Sarikaya <[e-mail]>]

Hi Dave,
Thanks for the notes.
A few corrections to the minutes will be issued soon.

Regards,

Behcet

On Tue, Mar 28, 2017 at 1:04 PM, David Allan I <[e-mail]>
wrote:

> Here are the notes from today’s meeting….
>
> Enjoy
>
> D
>
>
>
> *3/28/17 5gangup Side meeting*
>
> Behcet and Dirk
>
> ~25 in the room
>
> *Agenda bash*
>
> *5G IP issues – draft-vonhugo-5gangip-ip-issues-03*
>
> Current solution space consists of ILA and MAMS approaches.
>
> Incubating groups, time to incubate ourselves.
>
> Suggest it may be time to consider a BOF.
>
> Want to stay in the INTAREA.  Interest in the AMF, SMF, UPF and some
> aspects of the AN in the context of IP.
>
> Proposals for access, mobility, and session management.
>
> Dave: you described nodes, and we are a protocol shop. Are you referring
> to N1, N2. N3. N4 and N11 interfaces? Yes
>
> Dave: What are we going to do that 3GPP is not as this is their
> architecture, are these alternative protocols?
>
> Georg Mayer: It is the job of 3GPP and the CT group to decide protocols.
> For N3 and N2 we had discussions in the plenary. 3GPP will specify this,
> mainly in the RAN group to allow access agnostic connections. Not sure what
> protocols we will reuse from 4G. Not clear to my understanding what role
> IETF will play.  What I saw in the slicing side meeting gives me hope that
> coordination can be achieved.  3GPP can do the protocol work, and if IETF
> does not have suitable protocols we can adopt then we will specify them.
>
> Behcet: working on a network that can be build from scratch is of
> interest, IoT etc.
>
> Charter items….
>
> Operator interest items…
>
> Connectivity issues, centered around I-L split gaps.  ILA, ILNP, LISP….
>
> *ILA mobility use cases  [e-mail] <[e-mail]>*
>
> *V2I use case:*
>
> Asking INTAREA be taken up as a WG item.
>
> A number of requirements. Centered around security and dependency on the
> infrastructure.
>
> Road side unit (RSU) does ID translation?
>
> Dave: ILA’s claim to fame was no host stack change. With both LOC and ID
> Xlate this seems impossible.  A: (precis) Needs some thought but there
> would be ways to do this.
>
> Lorenzo:  <first question lost>
>
> Vehicle as only one IPv6 at a time.  It could have a subnet of
> identifiers.  Able to form IP addresses without coordination with the
> network. Behcet, take to the list.
>
> You would still need DHCP to give you an address or subnet.  You need more
> than one address, there is a best practice.   Read 7934.   Dave: PD
> wouldn’t work. Lorenzo: Correct you have to give all the bits to the
> network.
>
> Erik: You’ve reinvented IPv6 NAT.  Tom: Do we need to do extra encap.
> Erik: That was fine for the DC case. But if for a bunch of hosts inside the
> vehicle.  It is only the routing system that knows the ILA NAT (???)
>
> *Overview of MAMs:  Hannu Flinck*
>
> Framework for selecting access and core network independently, based on
> traffic types, changing network conditions, etc.  includes negotiation.
>
> *Draft-herbert-ila, and draft Mueller-ila-mobility*
>
> Erik: Is there a document outlining the problems with existing 3GPP
> protocols like GTP.  Behcet: pointed to the von-hugo draft.  Erik: AFAIK
> ILA for mobility is still a tunneling solution.
>
> *Draft-zhu-intarea-mams-control/user-protocol.*
>
> Georg: Issues draft would be of interest to 3GPP.  We are starting a 6
> month study in CT, which is also where the decisions will be taken,
> including w.r.t GTP.  I cannot judge this on a technical perspective, but
> you should bring the issues to 3GPP. Are you planning on doing such a
> thing. My expectation is that drafts take longer than 3GPP will take in
> making decisions.  If you send a whole document, there is a high
> probability of an ack but few reading it.
>
> Marco: I think it is a valid point to investigate shortcomings, but the CT
> work is DP and or CP?
>
> Georg: My understanding is all WG chairs we welcoming input but the window
> is short.  Do not want to be extreme or threatening, but the moment is now
> to provide input to 3GPP.
>
> Rudeiger (DT):  are you discussing a special release or a number of
> releases?  Georg: Release 15 will provide the foundation for the 5G system.
> What comes into the normative part now will not be easy to change in the
> coming releases. Introducing new concepts will become hard, such as
> replacing GTP.  Note that there might be multiple soluitions.
>
> Kalyani (Vz): There is a desire to have an alternative to GTP, we had PMIP
> in 4G. So an overlay plus a native routing option.  Such work would belong
> to INTAREA. Not sure how we can talk about functions and protocols here
> while 3GPP is working on them. <incomplete capture>
>
> Lorenzo: My understanding is 3GPP assigns a /64 to a device as part of a
> PDU context.  ILA cannot provide … If 3GPP cannot accept a regression down
> to a 128 per host, we would be contradicting our advice to 3GPP from 15
> years ago.
>
> Georg: You are working on technical solutions, that is great. If you could
> highlight the problems you see to 3GPP that would be great. Just a warning,
> if we work on issues in parallel, we will get decoupled.  If you could make
> 3GPP aware of alternative solutions that would be great.
>
>
>
> _______________________________________________
> 5gangip mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/5gangip
>
>