Re: [DMM] Comments and questions to "Architecture Discussion on SRv6 Mobile User plane" draft

Hi Hannu,

Thank you very much for your review and comments.

Please see in-line [MK].

On Mon, Apr 17, 2023 at 2:25 AM Hannu Flinck (Nokia) <hannu.flinck@nokia.com>
wrote:

> Hello
>
>
>
> Regarding https://datatracker.ietf.org/doc/draft-kohno-dmm-srv6mob-arch/
> please find my comments and feedback.
>
> Clearly deep discussion is needed to better understand the justifications.
>
>
>
> For section 2
>
>
>
> The claimed limitations should be spelled out for:
>
>
>
> Non-optimal for any-to-any communication
>
> Non-optimal for edge/distributed computing
>
> Non-optimal for fixed and mobile convergence (FMC)
>
>
>
> As you know 5G is supporting all of the abovementioned. What is the real
> issue that is alluded? Can these be qualified?
>

[MK]

The draft discusses the solution approach and its architectural benefits of
translating mobile session information into routing information.

Mobile session information is a function of M,N (GTP start point and end
point), whereas routing information is a function of N (destination).

So, for any-to-any communication, it's obvious. O(N^2) vs O(N).

Edge/distributed computing can be seen as a subset of any-to-any
communication. Routing paradigm naturally enables ubiquitous computing.

But Session based architecture requires session signaling for Edge Server
Selection and UPF Selection. 3GPP TS 23.548 and TS23.558 seem complex.
For FMC/WWC, 3GPP architecture requires that non-3GPP traffic are also to
be terminated by UPF, which is not optimal from the viewpoint of cost and
scaling.

>
> Regarding “No control of the underlay path” is by intention left out as a
> matter of implementation and deployment as the underlay can be of any
> technology and topology.
>
>
>
> Statement:
>
>    The IP routing paradigm naturally eliminates these tunnel session
>
>    based shortcomings.  Segment Routing enables fast protection, policy,
>
>    slicing, etc. to provide reliability and SLA differentiation.
>
>
>
> Because the shortcomings have not be opened up this statement of use of
> routing is without any concretism. Yes, segment routing enables the
> mentioned capabilities, but how do they match to the mentioned limitations
> is not clear. What is meant by routing should defined better. Do you define
> routing as SRv6 encapsulation only or something more?
>

[MK]
We will try to revise the writing to be more clear and concrete. Thank you
for pointing this out.

>  Furthermore, if routing protocols are used for mobility my question is
> the feasibility of seamless HO and how is the subnetting and address
> allocation arranged?
>

[MK]
We fully respect the current way of mobility. There are certainly routing
based shortcomings.  We have to experiment and verify the feasible
frequency of HO and how far it can be seamless. But mobile applications are
getting various, e.g. FWA, IoT..  And it would be good to apply this
architecture in a complementary and selective way.

Best regards,
Miya

>
>
>
> Best regards
>
> Hannu
>
>
> _______________________________________________
> dmm mailing list
> dmm@ietf.org
> https://www.ietf.org/mailman/listinfo/dmm
>