+ SPRING WG

Hi Darren,

Thanks for your review and feedback. Especially thanks for "looking beyond"
just this document.

Since this review is somewhat late (but very welcome), I would like to
clarify some aspects so it is brought to the attention of the appropriate
WGs and author teams.

On Thu, Jul 23, 2026 at 5:55 AM Darren Dukes via Datatracker <
[email protected]> wrote:

> Document: draft-ietf-idr-sr-policy-nrp
> Title: BGP SR Policy Extensions for Network Resource Partition
> Reviewer: Darren Dukes
> Review result: Ready with Issues
>
> I was asked to review revision 11 of this draft as part of the IntArea
> directorate. Since then revision 13 has been posted and I've revised this
> review to that version. This review is for the Int AD IESG review and may
> be
> considered by authors as last call review comments.
>
> General Observations:
> The specification is compact and the wire encoding is straightforward.
> The encoding and the separation between BGP validation and SR Policy
> Module (SRPM) semantic processing are consistent with RFC 9830.
>
> After reviewing the draft and its normative references I'm left with one
> ISSUE around validity checking.
>
> This draft defines the NRP ID Sub TLV and its encoding, but
> does not specify, nor fully abdicate, the validity checking of the NRP ID
> to
> draft-ietf-spring-sr-policy-nrp.
>
> draft-ietf-spring-sr-policy-nrp-02 does not yet define
> deterministic behavior when a syntactically valid NRP association cannot be
> used by the receiving head end.
>

KT> This is for the SPRING WG to resolve as it pertains to a SPRING WG
draft. May I request the authors of that document to track this as an issue?


>
> Further more this draft contains select normative language about how NRP
> IDs are
> associated with candidate paths, but that text says essentially nothing
> normative.
>
> This draft states in Operational Considerations:
>     Although the updates to SR Policy architecture in
>    [I-D.ietf-spring-sr-policy-nrp] allow different candidate paths in
>    one SR Policy to be associated with different NRPs, in normal network
>    scenarios it is considered that no matter which candidate path is
>    active, the association between an SR Policy and NRP is consistent.
>    In such case all candidate paths of one SR Policy SHOULD be
>    associated with the same NRP.  The cases where different candidate
>    paths of an SR Policy are associated with different NRPs is valid but
>    NOT RECOMMENDED.
>
> This does not tell an implementor of this specification anything other than
> "anything other NRP ID 0 is valid".
>
> ISSUE 1: I suggest this text say something deterministic or is removed
> completely and replaced with text stating "NRP ID validation is out of
> scope
> and defined in draft-ietf-spring-sr-policy-nrp"
>

KT> This is very good suggestion and aligned with BGP SR Policy SAFI draft
where the semantic validation is left to the SRPM module. Request the
authors to consider this.


>
> ISSUE  2 (not an issue for this draft): In addition
> draft-ietf-spring-sr-policy-nrp should be updated at the earliest to
> specify
> normative validity checking of NRP IDs and the requirements around their
> permitted values in SR Policy candidate paths, and their lack of existence
> when
> NRP ID 0 may be received but the TLV is ignored.
>

KT> This is also a very good suggestion for the authors of the SPRING WG
document and I request that they take care of this at the earliest.

Thanks,
Ketan
_______________________________________________
spring mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to