Re: [Rtg-yang-coord] [netmod] rib-data-model and routing-cfg

["Acee Lindem (acee)" <[e-mail]>]

On 10/13/15, 12:30 PM, "Ladislav Lhotka" <[e-mail]> wrote:

>
>> On 13 Oct 2015, at 17:20, Acee Lindem (acee) <[e-mail]> wrote:
>> 
>> Hi Lada, NETMOD,
>> 
>> So I think we should move forward this ietf-rtg-cfg so that it can be
>> augmented and other work can move forward. We are still in disagreement
>> with respect to routing-instance/interface configuration.
>> 
>>    - We feel the IPv4/IPv6 interfaces should reference the
>> routing-instance in their config state. This is consistent with
>> draft-rtgyangdt-rtgwg-device-model-01.txt.
>>    - You feel that the routing-instance should have a list of leaf-ref’s
>> to the interface. You believe the leaf-ref provides a level of
>>validation
>> due to the fact that references can be confined to routing-instance
>> references. However, heretofore, no models are referencing the interface
>> leaf-refs in the list.
>
>True, these models (ietf-isis, for instance) use leafrefs with
>"if:interface-ref" type. However, such leafrefs are under-constrained
>because they can be configured to refer to:
>
>- interfaces of any layer, including physical interfaces, VLAN trunks etc.

Actually, putting the routing-instance reference in the IPv4 and IPv6
interface models (i.e., RFC 7277) constrains the reference to layer 3 more
effectively than the list of leaf-refs.

>
>- interfaces assigned to any routing instance.

But the list of leaf-refs doesn’t insure an IPv4 interface or IPv6
interface isn’t included by a single routing-instance.

>
>I believe in all these cases the choice has to be limited to (1) L3
>interfaces, and (2) belonging to "own" routing instance. These
>constraints will have to be checked in server code somehow - I would
>prefer to have them represented in the data model.
>
>But if nobody shares this concern with me, I am not going to block the
>document on this issue.

I’d also be interested if anyone shares this concern.

Thanks,
Acee 


>
>Lada 
>
>> 
>> Other than the Routing YANG Design Team having chosen the first option -
>> are there any other opinions?
>> 
>> Thanks,
>> Acee
>> 
>> On 10/9/15, 9:00 AM, "netmod on behalf of Acee Lindem (acee)"
>> <[e-mail] on behalf of [e-mail]> wrote:
>> 
>>> Hi Lada, 
>>> I2RS is not chartered to do the base models. There are other routing
>>> models that reference routing-cfg and even in-progress implementations.
>>> 
>>> On 10/9/15, 4:13 AM, "netmod on behalf of Ladislav Lhotka"
>>> <[e-mail] on behalf of [e-mail]> wrote:
>>> 
>>>> Hi,
>>>> 
>>>> I am sorry for cross-posting but I think it is high time to decide the
>>>> relationship between the data models in i2rs-rib-data-model and
>>>> netmod-routing-cfg I-Ds because they clearly target the same
>>>>management
>>>> data in a router. I can see three possible scenarios:
>>>> 
>>>> 1. The i2rs-rib module will be modified to augment
>>>> ietf-routing/ietf-ipv[46]-unicast-routing.
>>> 
>>> This would seem to be the obvious choice.
>>> 
>>>> 
>>>> 2. The scope of ietf-routing will be changed so that it would address
>>>> only host routing and simple routers. The modelling work for advanced
>>>> routers will be done elsewhere.
>>>> 
>>>> 3. The work on netmod-routing-cfg will be stopped.
>>> 
>>> A fourth option would be for me to take over ownership, move the work
>>>to
>>> the RTG WG, and we’d recruit some strong authors/reviewers from
>>>operators
>>> and other vendors (involving the ADs in selection).
>>> 
>>> Thanks,
>>> Acee 
>>> 
>>> 
>>>> 
>>>> Opinions?
>>>> 
>>>> Thanks, Lada
>>>> 
>>>> --
>>>> Ladislav Lhotka, CZ.NIC Labs
>>>> PGP Key ID: E74E8C0C
>>>> 
>>>> 
>>>> 
>>>> 
>>>> _______________________________________________
>>>> netmod mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/netmod
>>> 
>>> _______________________________________________
>>> netmod mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/netmod
>> 
>
>--
>Ladislav Lhotka, CZ.NIC Labs
>PGP Key ID: E74E8C0C
>
>
>
>