Hi all, > because: > > - the RFCs are formally ambiguous, dating way back to RFC1863 (which > is acknowledged to be experimental, but documented the practices going > back to the mid 1990s) > > - each IXP which has gone one way or the other has carefully > considered reasons for doing what they did > > - because of this, it will not be possible to reach consensus on > interpreting NO_EXPORT or passing it through (we have tried) > > - the authors of this ID believe that the only way to fix this problem > is to acknowledge that it cannot be fixed, then move on.
Alright, it seems an Internet-Draft is warranted to clarify or improve what route servers should do with certain well-known communities. I support solving that problem. The way I look at it is quite simple, let's assess how big the populations are that need to do work: Option A) An I-D that clarifies the behaviour of NO_EXPORT on a RS (and updates RFC1997 and RFC7947) Option B) An I-D that defines behaviour for a new flavor of NO_EXPORT (NO_EXPORT_VIA_RS) on a RS (and updates RFC1997 and RFC7947) Option A's impact: * A few Internet Exchange RS operators need to deploy a new routing policy. Option B's impact: * Every non-RFC7947 EBGP speaker needs to start scrubbing the new NO_EXPORT_VIA_RS community. * Every IXP RS operator needs to deploy a new configuration. * Every IXP RS participant needs to memorize that there is a new specific NO_EXPORT-ish community. In other words, I don't think it is fair to follow option B (draft-hilliard-grow-no-export-via-rs-00) because it causes the most work. For the route server operators that want to follow RFC 1997 to the letter: honoring NO_EXPORT on a RS is a fringe activity. If there really is a usecase to do so the IX should offer a locally significant community. Also realize that if RFC 1997 is updated according to option A, not honoring NO_EXPORT means you'll be compliant with the standards. Option A aligns with the principle of least astonishment. I look forward to draft-hilliard-grow-no-export-via-rs-01 which drops the definition of a new NO_EXPORT-ish community and simply defines the pragmatic, participant-friendly behaviour that RFC7947 BGP speakers should apply. Kind regards, Job _______________________________________________ GROW mailing list GROW@ietf.org https://www.ietf.org/mailman/listinfo/grow