Re: [5gangip] To initiate user-plane study work in 3GPP

[Satoru Matsushima <[e-mail]>]

Alex,

> 
> I understand then.  Thank you for the explanation.
> 
> I think there are some problems with this.  Multiple encapsulations lead
> to loss of efficiency.  This leads to lesser bandwidth.  For voice
> calls, this can mean less quality of voice, or longer connection
> establishment times.
> 

As Cameron replied in another thread, the header tax issue has been lesser while bandwidth increased. What I mention here is that complication to correlate those stacked layers each other to meet the requirements.

Connection establishment time issue seems different from that. It more likes a control-plane signaling procedure issue.


> For example, now in Singapour I experience very long connection
> establishment times on my phone with SIM from home.  My operator is an MVNO (Mobile Virtual Network Operator) whose transport depends on another operator in my home country.  Here is yet another operator.
> 
> On another hand, skype has same connection establishment time as when I
> am at home.

So yes, that is a kind of control-plane signaling procedure issue.

Cheers,
--satoru



> 
> Alex
> 
> Le 16/11/2017 à 07:36, Satoru Matsushima a écrit :
>> Alex,
>> Let us imagine a case where under VPN label is imposed by a transport
>> service operator who provides backhaul connectivity for multiple
>> mobile operators. And upper VPN label is imposed by a mobile operator
>> who need to create three or more VPNs for each mobile user-plane,
>> control-plane O&M and if any other purposes respectively even just
>> within the mobile operator for example.
>> Cheers, --satoru
>>> 2017/11/15 22:46、Alexandre Petrescu <[e-mail]>の
>>> メール:
>>> Satoru,
>>> I see two distinct "VPN label" layers.  I can agree with two IP
>>> layers in same stack, but two VPN layers too?
>>> Arent these too many layers?
>>> For the question of which one of the realities happened in some
>>> operator - I can ask locally about which of these layers (VPN
>>> label, LSP label, VLAN ID, SMAC, DMAC, VPN label, LSP label, SMAC,
>>> DMAC) are seen at the PGW.
>>> Alex
>>> Le 13/11/2017 à 17:54, Satoru Matsushima a écrit :
>>>> Alex, You’re right as others have admitted. But you can see more
>>>> stacked layers underneath of GTP/UDP/IPv4 Please take a look at
>>>> the attached picture which one of the realities happened in some
>>>> operator. Cheers, --satoru
>>>>> 2017/11/12 20:26、Alexandre Petrescu
>>>>> <[e-mail]
>>>>> <mailto:[e-mail]>>のメール:
>>>>> Le 19/10/2017 à 04:52, Satoru Matsushima a écrit :
>>>>>> Thank you Dirk, Yes that presentation made by George has much
>>>>>> encouraged ietf community. Other key facts which the
>>>>>> justification text mentioned are following: o IAB statement
>>>>>> on IPv6 https://www.iab.org/2016/11/07/iab-statement-on-ipv6/ o IPv6
>>>>>> deployment measurement http://www.worldipv6launch.org/measurements/ We can see big
>>>>>> number of adoption rate for cellular operators especially in
>>>>>> US. T-Mobile US and Verizon Wireless exceed 80%, 47.51% for
>>>>>> ATT Wireless, and 59% for Sprint Wireless.
>>>>> YEs, these networks carry much IPv6, but is it carried mostly
>>>>> in GTP/UDP/IPv4?
>>>>> Alex
>>>>>> Cheers, --satoru
>>>>>>> 2017/10/19 2:17、<[e-mail]
>>>>>>> <mailto:[e-mail]>> <Dirk.von-
>>>>>>> [e-mail] <mailto:[e-mail]>>のメール:
>>>>>>> Dear all, I completely agree that we (i.e. at IETF
>>>>>>> community) should start work on potential protocol or
>>>>>>> framework solutions for 3GPP - even before 5G requirements
>>>>>>> (in architecture documents) are fully specified which as
>>>>>>> Georg in Prague pointed out
>>>>>>> https://datatracker.ietf.org/meeting/99/materials/slides-99-edu-sessk-3gpp-ietf-collaboration-on-5g/
>>>>>>> is expected for end of this year. And I also think that a
>>>>>>> major role will play Identity / Locator Separated protocols
>>>>>>> like LISP, ILNP, ILA ... supporting multiple types of
>>>>>>> mobility by choosing correspondingly adapted configurations
>>>>>>> ... which would fit to service-tailored network slicing
>>>>>>> ideas. Thanks! Best Regards Dirk -----Original
>>>>>>> Message----- From: 5gangip
>>>>>>> [mailto:[e-mail]] On Behalf Of Satoru
>>>>>>> Matsushima Sent: Mittwoch, 18. Oktober 2017 11:11 To:
>>>>>>> [e-mail]; [e-mail];
>>>>>>> [e-mail]; <[e-mail]> Cc:
>>>>>>> [e-mail] Subject: Re: [5gangip] To initiate
>>>>>>> user-plane study work in 3GPP
>>>>>>> Behcet, David, and Uma,
>>>>>>> Thank you for your comments. Those are really informative.
>>>>>>> And yes I agree that we need to work with 3GPP. That’s why
>>>>>>> the WID out there.
>>>>>>> While there’s no clear facts on future networking in 5G, I
>>>>>>> wrote up the text in justification with facts which at
>>>>>>> least IPv6 adoption has been growth and IAB recommend SDOs
>>>>>>> to review existing standards work in IPv6-only
>>>>>>> environment.
>>>>>>> Cheers, --satoru
>>>>>>>> 2017/10/18 6:52、[e-mail]のメール:
>>>>>>>> Uma
>>>>>>>> I agree that X2 is a complication but this is only used
>>>>>>>> temporarily to provide a fast switch-over during a
>>>>>>>> mobility event.  In terms of end-to-end communication,
>>>>>>>> even during an X2 event, the topology and use of a tunnel
>>>>>>>> anchored at the centre remains.
>>>>>>>> The issue is not the protocol but the time the underlying
>>>>>>>> equipment takes to update the tunnel path.
>>>>>>>> At the moment (note – AT THE MOMENT!) the use-case which
>>>>>>>> requires the most expeditious switch-over and is least
>>>>>>>> tolerant to packet drops due to mobility events is VoLTE
>>>>>>>> and it survives very well in current S1 and X2 handover
>>>>>>>> systems.
>>>>>>>> I think this - “I am not sure who is “we”  (I would see
>>>>>>>> that as 3GPP).”   - may be where we are struggling!
>>>>>>>> Until we have a use-case which patently fails using the
>>>>>>>> current anchored GTP system, we are going to find it very
>>>>>>>> hard to have an alternative solution considered.
>>>>>>>> We need to work with 3GPP on this….
>>>>>>>> David
>>>>>>>> From: Uma Chunduri [mailto:[e-mail]] Sent:
>>>>>>>> 17 October 2017 14:39 To: Lake D Mr (PG/R - Elec
>>>>>>>> Electronic Eng) <[e-mail]>;
>>>>>>>> [e-mail]; [e-mail];
>>>>>>>> [e-mail] Cc: [e-mail] Subject:
>>>>>>>> RE: [5gangip] To initiate user-plane study work in 3GPP
>>>>>>>> David,
>>>>>>>> Agree mostly on what you said.  2 comments though..
>>>>>>>>> Mobility events are handled quickly and with minimal
>>>>>>>>> packet loss
>>>>>>>> I don’t think so. It’s an overhead for RAN mobility (X2
>>>>>>>> tunnels creation and additional control signalling for
>>>>>>>> the same) and similar complication for UPF mobility.
>>>>>>>>> However, for NEW 5G use-cases such as Ultra-Low Latency
>>>>>>>>> and edge-based services we should be considering
>>>>>>>>> whether GTP is the correct choice provided we can meet
>>>>>>>>> the commercial and regulatory requirements of the MNOs
>>>>>>>>> in terms of LI and charging.
>>>>>>>> I am not sure who is “we”  (I would see that as 3GPP).
>>>>>>>> But it’s good to put proposals  to show what layer 3
>>>>>>>> mobility entails -  like the one Bob mentioned below or
>>>>>>>> what Dino
>>>>>>>> indicatedhttps://www.ietf.org/mail-archive/web/5gangip/current/msg00571.html
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
> --
>>>>>>>> Uma C.
>>>>>>>> From: 5gangip [mailto:[e-mail]] On Behalf
>>>>>>>> Of [e-mail] Sent: Tuesday, October 17, 2017
>>>>>>>> 2:15 PM To: [e-mail]; [e-mail];
>>>>>>>> [e-mail] Cc: [e-mail] Subject:
>>>>>>>> Re: [5gangip] To initiate user-plane study work in 3GPP
>>>>>>>> Bob
>>>>>>>> This is the key statement for me:
>>>>>>>> “A study group would have to work out what services GTP
>>>>>>>> provides”
>>>>>>>> Both in this group and related efforts (e.g. IDEAS at
>>>>>>>> IETF, NGP at ETSI), we need to consider how it is that
>>>>>>>> mobile networks are built.
>>>>>>>> In terms of P2P, traffic on today’s mobile networks is
>>>>>>>> predominantly anchored at a central point, not P2P
>>>>>>>> despite the associated engineering inefficiencies.
>>>>>>>> There are very good reasons for this:
>>>>>>>> • It makes LI very easy to implement • It makes
>>>>>>>> association of traffic with user clear allowing simple
>>>>>>>> billing mechanisms • Mobility events are handled quickly
>>>>>>>> and with minimal packet loss
>>>>>>>> In terms of the major use-cases today (voice over LTE and
>>>>>>>> mobile broadband) as they move to 5G, I really don’t see
>>>>>>>> any need to change the underlying tunnelling protocol
>>>>>>>> because It Works.
>>>>>>>> However, for NEW 5G use-cases such as Ultra-Low Latency
>>>>>>>> and edge-based services we should be considering whether
>>>>>>>> GTP is the correct choice provided we can meet the
>>>>>>>> commercial and regulatory requirements of the MNOs in
>>>>>>>> terms of LI and charging.
>>>>>>>> David
>>>>>>>> From: 5gangip [mailto:[e-mail]] On Behalf
>>>>>>>> Of Robert Moskowitz Sent: 17 October 2017 05:52 To:
>>>>>>>> [e-mail]; Satoru Matsushima
>>>>>>>> <[e-mail]> Cc: [e-mail] Subject: Re: [5gangip] To initiate user-plane study work
>>>>>>>> in 3GPP
>>>>>>>> Behcet and Satoru,
>>>>>>>> Last year I did a study of GTPv1U and worked out a
>>>>>>>> different approach that would better support P2P
>>>>>>>> communications without the need of a home proxy agent.
>>>>>>>> You can see part of it in:
>>>>>>>> draft-moskowitz-hip-ipnhip-02.txt
>>>>>>>> But I learned that there is tremendous resistance to any
>>>>>>>> change away from GTP, even a partial change.  A study
>>>>>>>> group would have to work out what services GTP provides.
>>>>>>>> Which services it does well and which it does poorly.
>>>>>>>> Then how to better meet the service needs today and going
>>>>>>>> forward.
>>>>>>>> Bob
>>>>>>>> On 10/16/2017 04:12 PM, Behcet Sarikaya wrote: Hi
>>>>>>>> Satoru,
>>>>>>>> Thanks for the info.
>>>>>>>> Can you please explain the objective items in your WID?
>>>>>>>> It seems like according to TR 29.891 that you mentioned,
>>>>>>>> Release 15 user plane protocol is GTPv1U and control
>>>>>>>> plane protocol for SBA is HTTP.
>>>>>>>> Regards, Behcet
>>>>>>>> On Mon, Oct 16, 2017 at 4:13 AM, Satoru Matsushima
>>>>>>>> <[e-mail]> wrote: FYI. I’d submit a
>>>>>>>> contribution to propose a new study item which is to
>>>>>>>> initiate user-plane protocol study work in 3GPP CT4 WG.
>>>>>>>> http://www.3gpp.org/ftp/tsg_ct/WG4_protocollars_ex-CN4/TSGCT4_80_Kochi/Docs/C4-175098.zip
>>>>>>>> 
>>>>>>>> 
> Please take a look at the doc from above link. I'd appreciate any your support if you think it makes sense.
>>>>>>>> Cheers, --satoru _______________________________________________ 5gangip
>>>>>>>> mailing list [e-mail] https://www.ietf.org/mailman/listinfo/5gangip
>>>>>>>> _______________________________________________ 5gangip
>>>>>>>> mailing list [e-mail] https://www.ietf.org/mailman/listinfo/5gangip
>>>>>>> _______________________________________________ 5gangip
>>>>>>> mailing list [e-mail] https://www.ietf.org/mailman/listinfo/5gangip
>>>>>> _______________________________________________ 5gangip
>>>>>> mailing list [e-mail] <mailto:[e-mail]> https://www.ietf.org/mailman/listinfo/5gangip
>>>>> _______________________________________________ 5gangip mailing
>>>>> list [e-mail] <mailto:[e-mail]> https://www.ietf.org/mailman/listinfo/5gangip