Re: [Rtg-yang-coord] [Teas] [mpls] Generic LSP Yang

Dhruv,

On 03/06/2015 07:00 AM, Dhruv Dhody wrote:
> Hi Lou,
> 
>> -----Original Message-----
>> From: Teas [mailto:teas-bounces@ietf.org] On Behalf Of Lou Berger
>> Sent: 05 March 2015 22:48
>> To: Dhruv Dhody; mpls@ietf.org
>> Cc: Rtg-yang-coord@ietf.org; Zhangxian (Xian); TEAS WG
>> Subject: Re: [Teas] [mpls] Generic LSP Yang
>>
>> Hi Dhruv,
>>     Good stuff!  There's already some related work going on based on the TE
>> LSP drafts/work started in MPLS and now in the TEAS WG.  -- This was
>> mentioned at the TEAS interim too, see
>> http://www.ietf.org/proceedings/interim/2014/12/18/teas/minutes/minutes-
>> interim-2014-teas-1
>>
>> It looks like you don't completely overlap as you are also considering non-TE
>> LSPs.
> 
> [Dhruv]: Yes, our aim is to go one level up and see if we can have a
> generic LSP yang model including TE and non-TE, as well any signaling
> protocol or lack thereof (static, SR).
> 

Understood.  BTW non-signaled TE is still part of TE architecture and
the yang model work underway in TEAS. (TP, and to some degree PCE really
made this point clear.)

>>
>> So I guess a basic question to answer is what is the relationship of the
>> models of TE and non-TE LSPs. e.g,
>> - a generic LSP model with parallel TE and non-TE sub models
>> - a generic LSP model with protocol specific models
>> - a generic LSP model with some mix of the above (which I think you are
>> proposing)
>> - no generic LSP model, and separate  TE and non-TE sub models
>> - etc
> 
> [Dhruv]: Our aim is to explore and see if we have attributes common
> to all LSP irrespective of TE/non-TE; signaling protocol, Segment
> Routing and even PCEP to warrant such a generic LSP model.
> 

Sure. PCE driven TE LSPs are in scope of the generic TE YANG models
being developed.   PCEP protocol specifics belong to PCE.

> A possible overall relationship with the model is in the draft - 
>                  +---------------+
>                  | Generic LSPDB |
>                  |   ietf-lspdb  |
>                  +-------^-------+
>                          |
>          ---------------------------------------------
>         |                |                |           |
>         |                |                |           |
>    +----+----+   +-------+-------+   +----+----+   +--+--+
>    | LDP-LSP |   |    BGP-LSP    |   |  TE-LSP |   | SR  |
>    |         |   |               |   |         |   |     |
>    +---------+   +---------------+   +----^----+   +--^--+
>                                           |           |
>                                            -----------
>                                           |           |
>                                           |           |
>                                      +----+----+   +--+--+
>                                      | RSVP-TE |   | SR- |
>                                      |         |   | TE  |
>                                      +----^----+   +--^--+
>                                           |           |
>                                           |           |
>                                            -----------
>                                           |
>                                      +----+----+
>                                      |  PCEP   |
>                                      |         |
>                                      +---------+
> 
> 
>>
>> It's not clear to me how much value a generic model brings, but details
>> always help to clarify the situation.
>>
>> My preference would be to have more details on the non-TE models (to compare
>> against the TE model) before deciding on the need for / value of a generic
>> LSP model.
> 
> [Dhruv]: Yes, I think that would surely help. 

Glad to hear we're in agreement.

> IMHO the ease of augmentation and having a generic base model is one
> of the key advantages of YANG, so we hope its a worthwhile exercise
> to explore the need for such a generic model instead of separate TE
> and non-TE models.
> 
The details of each will certainly demonstrate the degree of overlap.

Thanks,
Lou

> Regards,
> Dhruv
> 
>>
>> Lou
>>
>> On 3/5/2015 11:48 AM, Dhruv Dhody wrote:
>>> Hi,
>>>
>>> Xian and I have created generic base yang model for LSPs. This model
>>> is expected to be augmented by seperate data model with specific
>>> signalling protocol and technology.
>>>
>>> See https://tools.ietf.org/html/draft-zhang-mpls-lspdb-yang-00
>>>
>>> Comments / Suggestions ?
>>>
>>> We need to work out the relationship with the rest of MPLS yang model
>>> which can be hashed out in Dallas.
>>>
>>> Regards,
>>> Dhruv / Xian
>>>
>>> Excuse the obvious nit with the LSP abbreviation expansion :P
>>>
>>> _______________________________________________
>>> mpls mailing list
>>> mpls@ietf.org
>>> https://www.ietf.org/mailman/listinfo/mpls
>>>
>>
>>
>>
>> _______________________________________________
>> Teas mailing list
>> Teas@ietf.org
>> https://www.ietf.org/mailman/listinfo/teas
> 