Re: [Idr] Benjamin Kaduk's No Objection on draft-ietf-idr-bgp-optimal-route-reflection-25: (with COMMENT)

[Alvaro Retana <[e-mail]>]

Robert:

Hi!


Given the text in §4:

   The achievement of optimal routing between clients of different
   clusters relies upon all route reflectors learning all paths that are
   eligible for consideration.  In order to satisfy this requirement,
   BGP add-path [RFC7911] needs to be deployed between route reflectors.

...making rfc7911 a Normative reference makes sense to me.

Thanks!

Alvaro.

On June 16, 2021 at 5:08:03 PM, Robert Raszuk ([e-mail]) wrote:

Hello Ben & Alvaro,
>
> > Regarding moving referencing of RFC7911 to normative let's observe that
> it
> > may be required only in one specific deployment scenario of using
> multiple
> > clusters in non encapsulating networks for IPv4/IPv6 AFs. It is clearly
> not
> > required for other deployment models of BGP ORR. Therefore I am not sure
> if
> > it deserves the elevated level of referencing in the document. But if the
> > above use is a sufficient reason to treat it as normative reference I
> > personally have no problem moving it.
>
> I'm happy to leave this to the responsible AD; my comment stems from the
> IESG statement at
>
> https://www.ietf.org/about/groups/iesg/statements/normative-informative-references/
> that points out that "even references that are relevant only for optional
> protocol features must be classified as normative" (if they would otherwise
> meet the criteria.  Only needed in some deployment scenarios is not quite
> the same as being an optional feature, though, so I think there's still
> some element of judgment needed.


In the case of this work the situation is not really about Add-Paths RFC
being necessary for the proposed enhancement/protocol feature either
mandatory or optional. It is helpful to some Route Reflector deployment
scenarios with or without this extension. ORR only observes that to get
consistent IBGP paths Add-Paths may be used between RR clusters. So could
Diverse-Path technique and in some cases perhaps also consistent controller
based path installation. There can be even more ways to accomplish
consistent path selection pool across clusters.

But as you suggest I think moving RFC7911 to the normative reference should
not be a problem if AD or anyone else finds this appropriate.

I will therefore hold publishing -26 till we get some clarity on this point
from an AD.

Many thx,
Robert