[GROW] bmp loc-rib monitoring scope question (was Re: I-D Action: draft-ietf-grow-bmp-local-rib-02.txt)

[Jeffrey Haas <[e-mail]>]

On Mon, Sep 17, 2018 at 01:00:22PM -0700, [e-mail] wrote:
>         Title           : Support for Local RIB in BGP Monitoring Protocol (BMP)

This question is motivated by an implementation in-progress:

In section 5, we have the following text:

:   Loc-RIB contains all routes from BGP peers as well as any and all
:   routes redistributed or otherwise locally originated.  In this
:   context, only the BGP instance Loc-RIB is included.  Routes from
:   other routing protocols that have not been redistributed, originated
:   by or into BGP, or received via Adj-RIB-In are not considered.

The second sentence is slightly confusing to me, and perhaps the usual
headache about the definition of loc-rib in RFC 4271, redistribution
(section 9.4 of same) and how other implementations work.

Somewhat colloquially, BGP implementations tend to concern themselves with
distribution of the "active" routes in most cases.  Thus, the routes that
get sent are the ones that are in the Routing Table in the RFC 4271 sense.
While there are scenarios where routes that are not the active route are
sent (e.g. add-paths, best-external, etc.), typically the active route is
sent.  This does interact with policy.

Policy normally will say "redistribute other protocol" in the circumstance
where for a given destination the best route is not BGP, but a BGP route is
present.

So... what exactly is intended by that second sentence?  It seems to read
"only pay attention to BGP routes".  

Based on the primary use case for loc-rib (avoid the need for a parallel BGP
session to your BMP rib-in session), I suspect what's intended is "send the
route that's eligible to be sent to BGP".

-- Jeff