Hi Anoop, Note that it is not only the ESI-based filtering procedure (for multi-homing) that is performed using the information in the Ethernet option TLV, but also the E-Tree flags and the BUM indication. So even if you are using local-bias, and you are not using E-Tree, we still want the BUM indication to avoid transient packet duplication in certain scenarios.
So IMO what we want is to RECOMMEND the use of the ethernet option TLV in all cases. Thanks. Jorge From: Anoop Ghanwani <an...@alumni.duke.edu> Date: Sunday, December 18, 2022 at 3:49 AM To: Boutros, Sami <sbout...@ciena.com> Cc: Boutros, Sami <sboutros=40ciena....@dmarc.ietf.org>, UTTARO, JAMES <ju1...@att.com>, Jorge Rabadan (Nokia) <jorge.raba...@nokia.com>, Matthew Bocci (Nokia) <matthew.bo...@nokia.com>, bess@ietf.org <bess@ietf.org>, draft-ietf-bess-evpn-gen...@ietf.org <draft-ietf-bess-evpn-gen...@ietf.org> Subject: Re: [bess] [**EXTERNAL**] RE: CORRECTION WG Last Call, IPR and Implementation Poll for *draft-ietf-bess-evpn-geneve-03* Here's a proposed change: OLD While "local-bias" MUST be supported along with Geneve encapsulation, the use of the Ethernet option-TLV is RECOMMENDED to follow the same procedures used by EVPN MPLS. NEW While "local-bias" MUST be supported along with Geneve encapsulation, the use of the Ethernet option-TLV is required if not using local-bias and instead following procedures used by EVPN MPLS. This allows implementations that implement only local bias to skip sending/implementing the TLV. Or are you trying to RECOMMEND that implementations also implement the EVPN MPLS procedures in addition to local bias? If so, I would word the text as: While "local-bias" MUST be supported along with Geneve encapsulation and that doesn't require the use of the Ethernet option-TLV, it is RECOMMENDED that Geneve implementations also implement the same procedures used by EVPN MPLS. On Sat, Dec 17, 2022 at 6:33 PM Anoop Ghanwani <an...@alumni.duke.edu<mailto:an...@alumni.duke.edu>> wrote: Sami, Why is it recommended to carry the TLV if local bias is in use? (I understand the need for it if we're not using local bias.) Anoop On Sat, Dec 17, 2022 at 2:06 PM Boutros, Sami <sbout...@ciena.com<mailto:sbout...@ciena.com>> wrote: I’m not sure what tightening you are recommending, I am out of ideas of how to tighten this, may be you can propose something. It is quite clear to me and to the authors, and I hope to everyone else, how the TLV can be used for SH as a mechanism similar to local bias, as well it can be used when ETREE support is needed. Thanks, Sami From: Anoop Ghanwani <an...@alumni.duke.edu<mailto:an...@alumni.duke.edu>> Date: Saturday, December 17, 2022 at 1:25 PM To: Boutros, Sami <sbout...@ciena.com<mailto:sbout...@ciena.com>> Cc: Boutros, Sami <sboutros=40ciena....@dmarc.ietf.org<mailto:40ciena....@dmarc.ietf.org>>, UTTARO, JAMES <ju1...@att.com<mailto:ju1...@att.com>>, Rabadan, Jorge (Nokia - US/Mountain View) <jorge.raba...@nokia.com<mailto:jorge.raba...@nokia.com>>, Bocci, Matthew (Nokia - GB) <matthew.bo...@nokia.com<mailto:matthew.bo...@nokia.com>>, bess@ietf.org<mailto:bess@ietf.org> <bess@ietf.org<mailto:bess@ietf.org>>, draft-ietf-bess-evpn-gen...@ietf.org<mailto:draft-ietf-bess-evpn-gen...@ietf.org> <draft-ietf-bess-evpn-gen...@ietf.org<mailto:draft-ietf-bess-evpn-gen...@ietf.org>> Subject: Re: [bess] [**EXTERNAL**] RE: CORRECTION WG Last Call, IPR and Implementation Poll for *draft-ietf-bess-evpn-geneve-03* Sami, I don't believe there was closure on this issue. I think the text around the option TLV being RECOMMENDED should be tightened so that it's recommended only when needed. The way the draft is currently written, it sounds like it is recommending that the TLV always be carried if multihoming is in use. But this is not necessary or even valuable if Local Bias is in use. On Wed, Jul 13, 2022 at 12:12 AM Anoop Ghanwani <an...@alumni.duke.edu<mailto:an...@alumni.duke.edu>> wrote: > Hi Sami, > > Thanks for updating the doc. > > Regarding this: > >>> > > I find this statement confusing > > > > While "local-bias" MUST be supported along with GENEVE encapsulation, > > the use of the Ethernet option-TLV is RECOMMENDED to follow the same > > procedures used by EVPN MPLS. > > > > > > I'm not sure how it helps to carry an extra TLV when it is known > > that its absence or presence results in identical behavior. > > > > Sami: The new TLV is not there only for local bias! It is there for bum and > leaf/root indications too. So, we can’t simply not carry it. As for the text > above, we are saying setting the ESI label in the TLV will allow us to follow > the same Split horizon behavior of MPLS-EVPN with no need for local bias. It > is true local bias must be supported but this mechanism will work too if > available given that it is optional. > > >>> > > I'm not sure I follow what you're saying. The new TLV is actually not > needed for the Local Bias case because we already know how to make that > case work without it. It is, however, needed for the non-Local Bias case
_______________________________________________ BESS mailing list BESS@ietf.org https://www.ietf.org/mailman/listinfo/bess
