Friday, November 12, 2021< ^ >
Meetecho has set the subject to: v6ops IETF 112
Room Configuration
Room Occupants

[11:30:10] Meetecho joins the room
[11:35:04] alexamirante joins the room
[11:45:07] Bob Hinden_web_704 joins the room
[11:45:07] Alessandro Amirante_web_992 joins the room
[11:45:07] Daniam Henriques_web_630 joins the room
[11:45:07] Ole Trøan_web_884 joins the room
[11:45:07] Steffen Klassert_web_916 joins the room
[11:45:07] Gorry Fairhurst_web_638 joins the room
[11:45:38] <Ole Trøan_web_884> Dear Meetecho... mind promoting Bob and I to chairs together with Ron and Fred?
[11:45:39] Haibo Wang_web_335 joins the room
[11:45:45] abdullahalshoaili_web_636 joins the room
[11:45:46] <Alessandro Amirante_web_992> on it
[11:46:00] Bob Hinden_web_704 leaves the room
[11:46:04] Bob Hinden_web_252 joins the room
[11:46:10] Ole Trøan_web_884 leaves the room
[11:46:14] Ole Trøan_web_853 joins the room
[11:46:21] <Ole Trøan_web_853> Superb, thank you.
[11:47:28] Giuseppe Fioccola_web_457 joins the room
[11:47:29] <Alessandro Amirante_web_992> will you preload slide decks, or you want me to take care of that?
[11:47:40] <Ole Trøan_web_853> I am preloading now.
[11:47:48] <Alessandro Amirante_web_992> ack
[11:48:06] Jen Linkova_web_577 joins the room
[11:48:10] Jinyou Dai_web_363 joins the room
[11:49:15] Yisong Liu_web_130 joins the room
[11:49:47] <Bob Hinden_web_252> Hello 6MAN and V6OPS
[11:50:01] <Jen Linkova_web_577> Hi Bob
[11:50:16] <Bob Hinden_web_252> Hi Jen
[11:50:41] Xiao Min_web_866 joins the room
[11:50:51] Barbara Stark_web_218 joins the room
[11:51:37] Guozhen Dong_web_779 joins the room
[11:51:46] Dhruv Dhody_web_231 joins the room
[11:52:21] Shunwan Zhuang_web_212 joins the room
[11:52:36] Shuping Peng_web_161 joins the room
[11:52:59] Ketan Talaulikar_web_611 joins the room
[11:53:05] Zheng Zhang_web_315 joins the room
[11:53:16] <Dhruv Dhody_web_231> Yes!
[11:53:23] Roland Bless_web_876 joins the room
[11:53:58] Bill Fenner_web_948 joins the room
[11:53:59] Chang Liu_web_210 joins the room
[11:54:02] Paolo Volpato_web_834 joins the room
[11:54:23] Ron Bonica_web_321 joins the room
[11:54:36] Loa Andersson_web_674 joins the room
[11:54:46] Jeffrey Haas_web_926 joins the room
[11:54:51] Mohamed Boucadair_web_831 joins the room
[11:55:00] Shuping Peng_web_161 leaves the room
[11:55:02] Justin Iurman_web_530 joins the room
[11:55:04] Shuping Peng_web_460 joins the room
[11:55:04] Adrian Farrel_web_338 joins the room
[11:55:10] xiebin_web_676 joins the room
[11:55:32] Cheng Li_web_723 joins the room
[11:55:37] Tomohiro Fujisaki_web_149 joins the room
[11:55:59] <Barbara Stark_web_218> My tea is steeping. Almost done.
[11:56:10] <Cheng Li_web_723> hello
[11:56:19] Michael Richardson_web_400 joins the room
[11:56:21] Jie Dong_web_989 joins the room
[11:56:28] Jie Dong_web_989 leaves the room
[11:56:32] Jie Dong_web_467 joins the room
[11:56:35] Shuping Peng_web_460 leaves the room
[11:56:39] Shuping Peng_web_402 joins the room
[11:56:50] Yuehua Wei_web_447 joins the room
[11:56:59] John Border_web_604 joins the room
[11:57:07] Tim Chown_web_530 joins the room
[11:57:21] Stefano Previdi_web_656 joins the room
[11:57:33] Aihua Liu_web_359 joins the room
[11:58:01] Martin Horneffer_web_622 joins the room
[11:58:03] Gyan Mishra_web_112 joins the room
[11:58:10] Tianji Jiang_web_191 joins the room
[11:58:13] <Alessandro Amirante_web_992> Fred hasn't even tried to connect today AFAICT
[11:58:33] Eduard V_web_746 joins the room
[11:58:42] Andrew Alston_web_231 joins the room
[11:58:49] Éric Vyncke_web_747 joins the room
[11:58:54] <Ole Trøan_web_853> Early morning in California I suppose.
[11:58:54] Chongfeng Xie_web_517 joins the room
[11:58:56] Jingrong Xie_web_180 joins the room
[11:58:59] Bhavit Shah_web_106 joins the room
[11:59:02] Ville Hallivuori_web_218 joins the room
[11:59:03] Ben Maddison_web_733 joins the room
[11:59:04] Liu Yao_web_468 joins the room
[11:59:04] Chenhao Ma_web_963 joins the room
[11:59:11] Warren Kumari_web_874 joins the room
[11:59:18] Donald Eastlake_web_579 joins the room
[11:59:24] Martin Vigoureux_web_298 joins the room
[11:59:35] Gang Yan_web_110 joins the room
[11:59:44] Yoshifumi Atarashi_web_118 joins the room
[11:59:51] Tom Hill_web_113 joins the room
[11:59:53] Timothy Carlin_web_359 joins the room
[11:59:55] Eric Rosenberg_web_646 joins the room
[12:00:03] Darren Dukes_web_280 joins the room
[12:00:16] <Jen Linkova_web_577> How do we know if you are excited or angry, Ole? Can you give us typing rhythm examples?
[12:00:20] Erik Kline_web_837 joins the room
[12:00:21] Chen Li_web_575 joins the room
[12:00:22] <Barbara Stark_web_218> And notes are ready to go... :)
[12:00:27] Tom Herbert_web_345 joins the room
[12:00:27] Torunn Narvestad_web_449 joins the room
[12:00:30] Pablo Camarillo_web_421 joins the room
[12:00:39] Steven Hartley_web_902 joins the room
[12:00:43] Geir Egeland_web_487 joins the room
[12:00:46] Wim Henderickx_web_606 joins the room
[12:00:48] Gunter Van de Velde_web_638 joins the room
[12:00:53] Yasunobu Toyota_web_520 joins the room
[12:00:55] Greg Mirsky_web_689 joins the room
[12:00:56] Jim Uttaro_web_528 joins the room
[12:00:58] Martin Vigoureux_web_298 leaves the room
[12:00:59] Marco Davids_web_406 joins the room
[12:01:12] Chongfeng Xie_web_517 leaves the room
[12:01:16] Chongfeng Xie_web_102 joins the room
[12:01:16] Martin Vigoureux_web_155 joins the room
[12:01:17] Balazs Varga_web_308 joins the room
[12:01:28] Stuart Cheshire_web_989 joins the room
[12:01:35] Michael zhou_web_534 joins the room
[12:01:36] Haoyu Song_web_436 joins the room
[12:01:38] David Pelton_web_577 joins the room
[12:01:50] Jeffrey Haas_web_926 leaves the room
[12:01:54] Jeffrey Haas_web_441 joins the room
[12:01:56] Yuji Tochio_web_561 joins the room
[12:02:05] Kireeti Kompella_web_138 joins the room
[12:02:06] Darren Dukes_web_280 leaves the room
[12:02:10] Darren Dukes_web_613 joins the room
[12:02:27] Pablo Camarillo_web_421 leaves the room
[12:02:30] Ville Hallivuori_web_218 leaves the room
[12:02:31] Pablo Camarillo_web_431 joins the room
[12:02:34] Ville Hallivuori_web_437 joins the room
[12:02:34] Daniel Huang_web_230 joins the room
[12:02:50] Chongfeng Xie_web_102 leaves the room
[12:02:54] Chongfeng Xie_web_277 joins the room
[12:02:58] Fred Baker_web_995 joins the room
[12:03:03] Pascal Thubert_web_135 joins the room
[12:03:05] Yuya Kawakami_web_504 joins the room
[12:03:15] Qin Wu_web_863 joins the room
[12:03:22] Fan Yang_web_793 joins the room
[12:03:33] Martin Vigoureux_web_155 leaves the room
[12:03:53] Antoine Fressancourt_web_885 joins the room
[12:04:05] Martin Vigoureux_web_320 joins the room
[12:04:11] Haojie Wang_web_203 joins the room
[12:04:28] Pablo Camarillo_web_431 leaves the room
[12:04:31] mackermann@bcbsm.com_web_126 joins the room
[12:04:32] Pablo Camarillo_web_900 joins the room
[12:04:34] Ana Custura_web_309 joins the room
[12:04:43] Ignas Bagdonas_web_490 joins the room
[12:04:44] Geraldo Júnior_web_482 joins the room
[12:04:49] <Cheng Li_web_723> the audio is broken, any one has the same issue?
[12:04:56] <Warren Kumari_web_874> Nope
[12:04:59] <Ole Trøan_web_853> Sounds good here
[12:05:02] Timothy Winters_web_155 joins the room
[12:05:04] Erik Nygren_web_250 joins the room
[12:05:05] <Jen Linkova_web_577> I can hear Ron loud and clear
[12:05:14] Chi-Jiun Su_web_266 joins the room
[12:05:21] <Kireeti Kompella_web_138> all good here too
[12:05:21] <Cheng Li_web_723> ok, then issue on my side, let me reconnect
[12:05:25] Cheng Li_web_723 leaves the room
[12:05:29] Cheng Li_web_212 joins the room
[12:05:54] Rob Hamilton_web_939 joins the room
[12:05:56] Chongfeng Xie_web_277 leaves the room
[12:06:00] Jiamin_web_620 joins the room
[12:06:00] Chongfeng Xie_web_350 joins the room
[12:06:15] <Warren Kumari_web_874> Nicely done slides...
[12:06:26] Jordi Palet Martinez_web_853 joins the room
[12:06:31] <Dhruv Dhody_web_231> Would wearing an ICANN tshirt to IETF be akin to wearing another bands merch to a concert :)
[12:06:46] Jordi Palet Martinez_web_853 leaves the room
[12:06:49] Xueyan Song_web_689 joins the room
[12:06:50] Jordi Palet Martinez_web_514 joins the room
[12:07:02] Martin Vigoureux_web_320 leaves the room
[12:07:03] Jeffrey Haas_web_441 leaves the room
[12:07:14] Paul Bottorff_web_796 joins the room
[12:07:24] Martin Vigoureux_web_435 joins the room
[12:07:32] Jurgen Sonnenmoser_web_167 joins the room
[12:07:35] Vishnu Beeram_web_373 joins the room
[12:07:48] <Jen Linkova_web_577> Dhruv: maybe not ICANN but if there is such thing as ITU t-shirt - yes ;)
[12:07:53] Jeffrey Haas_web_483 joins the room
[12:07:58] <Dhruv Dhody_web_231> :rolling_on_the_floor_laughing:
[12:08:03] Cheng Li_web_212 leaves the room
[12:08:07] Cheng Li_web_347 joins the room
[12:08:11] <Shuping Peng_web_402> Very good slides! @Ron
[12:08:13] Chen Li_web_575 leaves the room
[12:08:48] Jyrki Soini_web_174 joins the room
[12:08:48] Joao Damas_web_930 joins the room
[12:09:19] Mark Andrews_web_969 joins the room
[12:09:25] Zongpeng Du_web_858 joins the room
[12:09:26] Stewart Bryant_web_669 joins the room
[12:09:26] Nicholas Gajcowski_web_589 joins the room
[12:09:33] Darren Dukes_web_613 leaves the room
[12:09:35] Juan Cerezo_web_708 joins the room
[12:09:37] Darren Dukes_web_339 joins the room
[12:09:38] Martin Vigoureux_web_435 leaves the room
[12:09:42] Martin Vigoureux_web_232 joins the room
[12:09:55] Shuping Peng_web_402 leaves the room
[12:09:59] Shuping Peng_web_547 joins the room
[12:10:05] Quan Xiong_web_238 joins the room
[12:10:18] <Adrian Farrel_web_338> About now in this discussion, I like to remind y'all about RFC 6398 and its commentary on Router Alert
[12:10:25] Himanshu Shah_web_161 joins the room
[12:11:14] <Erik Kline_web_837> the note that in section 7, it seems
[12:11:17] <Erik Kline_web_837> they*
[12:11:25] Kiran Makhijani_web_814 joins the room
[12:11:26] Shuping Peng_web_547 leaves the room
[12:11:30] Shuping Peng_web_635 joins the room
[12:11:40] Rurick Kellermann_web_449 joins the room
[12:11:56] Geng-Da Tsai_web_850 joins the room
[12:11:57] Martin Vigoureux_web_232 leaves the room
[12:12:01] Martin Vigoureux_web_111 joins the room
[12:12:01] Nigel Davis_web_823 joins the room
[12:12:05] Sheng Liu_web_683 joins the room
[12:12:24] Ahmed Bashandy_web_107 joins the room
[12:12:33] <Ole Trøan_web_853> For MLD Router Altert is limited to directly connected nodes only. Is Router Alert with RSVP used across a multi-hop path?
[12:12:58] Weiqiang Cheng_web_162 joins the room
[12:13:02] Jeffrey Haas_web_483 leaves the room
[12:13:06] Jeffrey Haas_web_236 joins the room
[12:13:11] Ching-Heng Ku_web_671 joins the room
[12:13:34] Martin Vigoureux_web_111 leaves the room
[12:13:36] James Evans_web_181 joins the room
[12:13:38] Martin Vigoureux_web_931 joins the room
[12:13:40] <Éric Vyncke_web_747> Ole, this is indeed my thinking, 1/ MLD are local 2/ RSVP is anyway limited to a single 'limited' domain for years
[12:13:42] Tong Zhu_web_349 joins the room
[12:13:47] Jeffrey Haas_web_236 leaves the room
[12:13:51] Jeffrey Haas_web_554 joins the room
[12:13:51] <Eduard V_web_746> IMHO About HbH: the problem is just an absence for the business case because processing additional headers are expenses and additional security risks. The business case would be found in closed domains sooner or later. It would probably not happen on the wide Internet. Not much IETF could do about it. Hence, even the current state is fine.
[12:14:05] <Ron Bonica_web_321> I don't know of any RSVP/IPv6 implementations. Most IPv4 implementations don't use Router Alert any more
[12:14:12] <Jen Linkova_web_577> Ole: is RSVP over IPv6 a thing?
[12:14:28] <Éric Vyncke_web_747> Jen: is RSVP over IPv4 (for QoS) a thing ?
[12:14:35] Erik Nygren_web_250 leaves the room
[12:14:39] Erik Nygren_web_160 joins the room
[12:14:50] <Jen Linkova_web_577> Eric: who said anything about QoS? I was thinking about RSVP-TE and CSPF
[12:14:51] <Adrian Farrel_web_338> I know of IPv6 RSVP-TE (not RSVP) and it does not use RA
[12:14:54] <Ole Trøan_web_853> Eduard: Business case -> Any use case?
[12:14:56] Jiamin_web_620 leaves the room
[12:14:58] <Shuping Peng_web_635> A few new cases that are using HBH will be presented soon :) @ Eduard
[12:15:00] Jiamin_web_389 joins the room
[12:15:05] <Ron Bonica_web_321> @Eric I don't know of any RSVP/Intserv implementations
[12:15:21] <Éric Vyncke_web_747> @Ron me neither ;-)
[12:15:28] <Kireeti Kompella_web_138> rsvp-te allows setting the dest IP to the next hop, so doesn't need router alert any more
[12:15:29] <Jen Linkova_web_577> Adrian: oh cool, good to know!
[12:15:31] Xueyan Song_web_689 leaves the room
[12:15:35] Xueyan Song_web_132 joins the room
[12:15:36] Ana Custura_web_309 leaves the room
[12:15:39] Jeffrey Haas_web_554 leaves the room
[12:15:43] Jiamin_web_389 leaves the room
[12:15:47] Jiamin_web_479 joins the room
[12:15:50] Jiamin_web_479 leaves the room
[12:15:54] Jiamin_web_346 joins the room
[12:16:05] <Bill Fenner_web_948> Fun fact: Linux has a terrible bug in their IPv6 Router Alert implementation (as an endpoint) which will cause packets containing RA to be delivered across network namespaces with no indication of which namespace they belong to.
[12:16:09] <Eduard V_web_746> Ole, the business case is not the same as the use case. Use cease has no financial justification that is needed for business.
[12:16:11] Xia Chen_web_701 joins the room
[12:16:31] Guozhen Dong_web_779 leaves the room
[12:16:36] Jared Mauch_web_312 joins the room
[12:16:48] Xueyan Song_web_132 leaves the room
[12:16:50] Guozhen Dong_web_920 joins the room
[12:16:52] Xueyan Song_web_306 joins the room
[12:17:06] <Ole Trøan_web_853> Eduard: My point was that we don't have any Internet wide use case. Therefore it would also be hard to have a business case. Lots of interesting stuff happening in limited domains with HBH though.
[12:17:19] <Eduard V_web_746> Agree
[12:17:46] Gabriel Montenegro_web_335 joins the room
[12:17:49] Martin Vigoureux_web_931 leaves the room
[12:17:53] Martin Vigoureux_web_568 joins the room
[12:18:03] Xueyan Song_web_306 leaves the room
[12:18:07] Xueyan Song_web_580 joins the room
[12:18:23] Joao Damas_web_930 leaves the room
[12:18:27] Joao Damas_web_350 joins the room
[12:18:30] Jeffrey Haas_web_198 joins the room
[12:18:31] Jiamin_web_346 leaves the room
[12:18:52] <Ole Trøan_web_853> the only one I can think of that's Internet applicability is the PMTU record option. But that use case was described in 1083 or was it 1063
[12:19:03] Ville Hallivuori_web_437 leaves the room
[12:19:06] Guozhen Dong_web_920 leaves the room
[12:19:07] Ville Hallivuori_web_531 joins the room
[12:19:10] Guozhen Dong_web_991 joins the room
[12:20:09] Jiamin_web_486 joins the room
[12:20:13] Jiamin_web_486 leaves the room
[12:20:16] Jiamin_web_634 joins the room
[12:20:23] <Eduard V_web_746> Even PMTU would not be accepted on the wide scale of the Internet. Why spend more money on headers processing if 97% of subscribers do not agree to pay a premium.
[12:20:40] <Fred Baker_web_995> Ron, I know of an RSVP/Interserv implementation. I wrote it. It is now somewhat historical.
[12:21:08] <Éric Vyncke_web_747> Curious to see how other modern routers (i.e., purely virtual routers in a containers/VM network) behave wrt HbH
[12:21:22] Guozhen Dong_web_991 leaves the room
[12:21:26] Guozhen Dong_web_654 joins the room
[12:21:34] <Darren Dukes_web_339> Eduardo
[12:21:51] <Darren Dukes_web_339> @eduard +1
[12:22:12] <Ole Trøan_web_853> In VPP we have implementations for various IOAM features. Also the PMTU record HBH option. We are not happy about the L4 not being at a fixed offset in packet though.
[12:22:23] Xueyan Song_web_580 leaves the room
[12:22:27] Xueyan Song_web_657 joins the room
[12:23:11] <Tim Chown_web_530> ioam would be nice to have for some e2e cases
[12:23:39] Xueyan Song_web_657 leaves the room
[12:23:41] Guozhen Dong_web_654 leaves the room
[12:23:43] Jeffrey Haas_web_198 leaves the room
[12:23:43] Xueyan Song_web_578 joins the room
[12:23:45] Guozhen Dong_web_364 joins the room
[12:23:47] Jeffrey Haas_web_880 joins the room
[12:24:08] <Tim Chown_web_530> the equivalent int protocol is probably more efficient, using p4
[12:24:10] <Tom Herbert_web_345> Ole, L4 being readable by intermeidate nodes is a probalem in itsef.
[12:24:47] <Kireeti Kompella_web_138> some = just about all?
[12:24:55] <Tom Herbert_web_345> I don't believe RFC8200 advocates dropping packets with HBH.
[12:24:56] Xueyan Song_web_578 leaves the room
[12:25:00] Xueyan Song_web_153 joins the room
[12:25:18] <Tim Chown_web_530> indeed there are many p4 implementations doing "hbh" behaviour.
[12:25:20] <Eduard V_web_746> iOAM is desperately needed because the firmware is becoming more and more buggy. The control plane does not understand that the data plane is dropping traffic in reality. Hence, closed domains should adopt it very soon.
[12:25:24] Zafar Ali_web_772 joins the room
[12:25:37] <Jen Linkova_web_577> @Tom: I'm afraid it's like saying that the force of gravity is a problem in itself
[12:26:12] Xueyan Song_web_153 leaves the room
[12:26:16] Xueyan Song_web_437 joins the room
[12:26:48] Steffen Klassert_web_916 leaves the room
[12:26:52] Steffen Klassert_web_204 joins the room
[12:26:52] Guozhen Dong_web_364 leaves the room
[12:26:56] Guozhen Dong_web_435 joins the room
[12:27:00] chenjinqiang_web_671 joins the room
[12:27:26] Zongpeng Du_web_858 leaves the room
[12:27:29] Xueyan Song_web_437 leaves the room
[12:27:30] Zongpeng Du_web_458 joins the room
[12:27:33] Xueyan Song_web_436 joins the room
[12:27:59] Martin Vigoureux_web_568 leaves the room
[12:28:03] Martin Vigoureux_web_914 joins the room
[12:28:10] Nigel Davis_web_823 leaves the room
[12:28:17] Guozhen Dong_web_435 leaves the room
[12:28:21] Guozhen Dong_web_152 joins the room
[12:28:43] Xueyan Song_web_436 leaves the room
[12:28:47] Xueyan Song_web_837 joins the room
[12:29:13] Zheng Zhang_web_315 leaves the room
[12:29:17] Jyrki Soini_web_174 leaves the room
[12:29:17] Zheng Zhang_web_629 joins the room
[12:29:21] <Darren Dukes_web_339> The v6 header is processed at each hop because transporting packets is always necessary.  Hey is not since it makes no money. In a limited domain it has a chance since it can bring enough value to justify the costs. I wonder if that acknowledgement would be useful?
[12:29:31] Guozhen Dong_web_152 leaves the room
[12:29:35] Guozhen Dong_web_588 joins the room
[12:30:15] <Darren Dukes_web_339> *hey=HBH :).
[12:30:37] Steffen Klassert_web_204 leaves the room
[12:30:41] Steffen Klassert_web_712 joins the room
[12:30:51] <Éric Vyncke_web_747> I was wondering about "Hey is not since" but now I understand it ;-)
[12:31:11] mcr joins the room
[12:31:46] Guozhen Dong_web_588 leaves the room
[12:31:50] Guozhen Dong_web_262 joins the room
[12:31:57] Nathalie Trenaman_web_751 joins the room
[12:32:02] <Tom Herbert_web_345> Is there data on the impact of RFC8200? Are any vendors ignoring HBH now instead of dropping?
[12:32:07] <Darren Dukes_web_339> Sorry iPad keyboard
[12:32:16] <Barbara Stark_web_218> Oh! Autocorrect of HbH to Hey
[12:32:31] <mcr> @Tom, I think that the problem is that we need 2-3 router lifetimes to expunge the pre-8200 routers.
[12:33:19] <Ole Trøan_web_853> And even so I don't think random end-user can inject packets that makes random SP's routers change their forwarding behaviour...
[12:33:24] <Cheng Li_web_347> i think within the limited domain, HBH can be used. ACL can be deployed at the edge nodes, which prevents the potential attacking from outside
[12:33:33] <Ole Trøan_web_853> So hard to see how most of HBH will not be limited to limited domain...
[12:33:54] Zheng Zhang_web_629 leaves the room
[12:33:58] Zheng Zhang_web_386 joins the room
[12:34:03] <Cheng Li_web_347> so that the sources of inserting HBH into the packets are controlled and trusted, then problem solve?
[12:34:05] joins the room
[12:34:26] <Eduard V_web_746> I do not see any problem now with the limited domain. iOMA or whatever - easy to implement.
[12:35:18] Steven Hartley_web_902 leaves the room
[12:35:22] Steven Hartley_web_519 joins the room
[12:35:25] <Cheng Li_web_347> the operators should say, let's upgrade all the devices in my network, then it can work
[12:35:26] <Jen Linkova_web_577> @mcr:  first when I read ' 2-3 router lifetimes to expunge the pre-8200 routers.' I was thinking 'so Router Lifetime is 16 bit integer, so..' ;)
[12:36:16] <mcr> @Jen. haha. half-life of utility of equipement is long due to movement to edge, and of course, ebay.
[12:36:18] Tong Zhu_web_349 leaves the room
[12:36:55] <Kireeti Kompella_web_138> just deprecate router alert
[12:37:23] <Ole Trøan_web_853> @Kireeti any suggestions for how to fix MLD?
[12:37:28] Zongpeng Du_web_458 leaves the room
[12:37:32] Zongpeng Du_web_862 joins the room
[12:37:34] <mcr> deprecate MLD :-)
[12:37:54] <Erik Kline_web_837> deprecate router alert with non-link-local communications?
[12:37:57] <Ole Trøan_web_853> @mcr: I've stopped sending MLD for link-local addresses, cause they don't add value anyway...
[12:38:14] <mcr> draft-ietf-6lo-multicast-registration for the win.
[12:39:01] <Cheng Li_web_347> It may can be deprecated, since young people don't know it actually(I guess), low value it appears.
[12:39:03] Guozhen Dong_web_262 leaves the room
[12:39:07] Guozhen Dong_web_479 joins the room
[12:39:24] <mcr> I'm wondering how many have read RFC9008, which is basically about how to use HBH in the edge, without pissing off the middle.
[12:40:34] Xueyan Song_web_837 leaves the room
[12:40:36] <Ole Trøan_web_853> @mcr: Can you give the one sentence summary?
[12:40:38] Xueyan Song_web_681 joins the room
[12:41:04] Shuai Zhang_web_486 joins the room
[12:41:11] Steven Hartley_web_519 leaves the room
[12:41:38] <Tom Herbert_web_345> Not sure something like IOAM will work with as a fixed length option
[12:41:42] Juhamatti Kuusisaari_web_389 joins the room
[12:41:49] Xueyan Song_web_681 leaves the room
[12:41:53] Xueyan Song_web_898 joins the room
[12:41:57] <Ole Trøan_web_853> @Tom scratch space like 791 record route?
[12:42:02] <mcr> RFC9008 explains how to add/subtract HBH options for LLNs that need to use source routing and loop detection.  RFC8200 allowed us to simplify things, as we can assume that it's okay for HBH to "escape" if the packet has to cross Internet.
[12:42:32] <Shuping Peng_web_635> for the ioam, we could separate the instruction and the data collected @ Tom
[12:42:34] <mcr> We have to use IPIP headers to add/subtract.  Fortunately, we know how to compress them.
[12:42:36] <Jared Mauch_web_312> (7 years is the reasonable lifetime of hardware, with some going much longer, expect the factory code may never change on the device)
[12:43:05] Xueyan Song_web_898 leaves the room
[12:43:09] Xueyan Song_web_679 joins the room
[12:43:39] Michael Breuer_web_400 joins the room
[12:44:17] Shuai Zhang_web_486 leaves the room
[12:44:21] Shuai Zhang_web_641 joins the room
[12:44:40] Steven Hartley_web_450 joins the room
[12:45:13] Mohamed Boucadair_web_831 leaves the room
[12:45:17] Mohamed Boucadair_web_238 joins the room
[12:45:23] Cheng Li_web_347 leaves the room
[12:45:27] Cheng Li_web_687 joins the room
[12:45:53] Zongpeng Du_web_862 leaves the room
[12:45:57] Zongpeng Du_web_252 joins the room
[12:46:14] Ran Chen_web_673 joins the room
[12:46:27] Rakesh Gandhi_web_702 joins the room
[12:46:50] Guozhen Dong_web_479 leaves the room
[12:46:54] Guozhen Dong_web_120 joins the room
[12:47:24] Radu Valceanu_web_642 joins the room
[12:47:29] <Darren Dukes_web_339> @bob I don’tthink there is practical internet wide use cases.
[12:47:47] <Jared Mauch_web_312> +1 {ddukes,bob}
[12:47:49] Jingrong Xie_web_180 leaves the room
[12:47:53] <Éric Vyncke_web_747> s/use cases/business cases/
[12:48:00] <Kireeti Kompella_web_138> we don't know if everyone drops hbh first and ask questions later
[12:48:05] Cheng Li_web_687 leaves the room
[12:48:09] Cheng Li_web_979 joins the room
[12:48:13] <Darren Dukes_web_339> @eric. Agreed
[12:48:28] Shuai Zhang_web_641 leaves the room
[12:48:40] <Bob Hinden_web_252> Use case drives the business case :-)
[12:48:50] Xueyan Song_web_679 leaves the room
[12:48:53] <Shuping Peng_web_635> agree @ Bob
[12:48:54] Xueyan Song_web_193 joins the room
[12:49:02] <Jared Mauch_web_312> (there's this interesting meta thing about how the internet is largely driven by business cases vs research)
[12:49:19] Nicholas Gajcowski_web_589 leaves the room
[12:49:42] Guozhen Dong_web_120 leaves the room
[12:49:50] Shuai Zhang_web_381 joins the room
[12:49:53] <Fred Baker_web_995> I thought that the IPv6 header had to fit in the first fragment? THat's a limit.
[12:50:02] <Adrian Farrel_web_338> (@jared and how we tend to squash new features based on "we don't see the business case" when in practice the actual business case for stuff is rarely what stuff was engineered for)
[12:50:12] Martin Vigoureux_web_914 leaves the room
[12:50:16] Martin Vigoureux_web_712 joins the room
[12:50:29] <Haoyu Song_web_436> no internet-wide use cases or it's more difficult to implement internet-wide use cases? I think the latter case is true.
[12:50:30] <Kireeti Kompella_web_138> number of ext hdrs and length thereof are easyish to specify. actual processing time (cycles) much harder
[12:50:38] <Jared Mauch_web_312> there's practically no fragmentation in v6
[12:50:51] Greg Mirsky_web_689 leaves the room
[12:50:55] Greg Mirsky_web_771 joins the room
[12:50:57] Himanshu Shah_web_161 leaves the room
[12:51:20] Srihari Sangli_web_758 joins the room
[12:51:22] <Bob Hinden_web_252> @Fred   Yes, IPv6 header, HBH headers, Dest EHs, and Transport need to be in the first fragment.
[12:51:25] <Jared Mauch_web_312> adrian: yes, often someone figures out how to [ab]use something later to get some goal
[12:51:30] <Jeffrey Haas_web_880> and dropping fragments is one of people's favorite mitigation rules for dealing with DDoS
[12:51:37] <Fred Baker_web_995> @Jared, agreed. That makes requiring the entire IPv6 header (including HBH and all other options) to be in the first packet constitutes a limit.
[12:51:49] <Éric Vyncke_web_747> @kireeti putting limit is "easy"  in 2021 but what will be the limit in 10 years ?
[12:52:01] <Kireeti Kompella_web_138> very true
[12:52:02] Shuai Zhang_web_381 leaves the room
[12:52:06] Shuai Zhang_web_454 joins the room
[12:52:17] <Jen Linkova_web_577> Jared: as someone who just recently has a very exciting troubleshooting session involving fragmentation, I disagree :))
[12:52:24] Mohamed Boucadair_web_238 leaves the room
[12:52:26] Guozhen Dong_web_645 joins the room
[12:52:27] <Bob Hinden_web_252> @Eric  That is the challenge to setting a limit
[12:52:28] Mohamed Boucadair_web_526 joins the room
[12:52:30] <Éric Vyncke_web_747> @Jen tell us more ;-)
[12:52:41] <Ole Trøan_web_853> We could also make the HBH header a terminating header (i.e. no headers following it).
[12:52:49] Himanshu Shah_web_526 joins the room
[12:53:16] <Fred Baker_web_995> @Ole, so a source-routed packet may not contain an HBH?
[12:53:19] <Jared Mauch_web_312> fred: as the practical limit of most devices these days is a pps one, not a bps one, any of these additional things (including uRPF, etc) often limit the pps, or halve the performance or more.
[12:53:25] <Ole Trøan_web_853> Jen: Didn't we agree to deprecate fragments.
[12:53:40] <Jen Linkova_web_577> Eric: there are protocols which are still want to fit the whole communication in a single datagram, you know ;)
[12:53:49] <Éric Vyncke_web_747> ;-)
[12:53:52] Dawei Fan_web_992 joins the room
[12:53:54] <Ole Trøan_web_853> Fred: The routing header does not use HBH. But yes, it would put some severe limitations on uses.
[12:54:02] <Éric Vyncke_web_747> @Jen; NFS over Ipv6 ?
[12:54:10] Guozhen Dong_web_645 leaves the room
[12:54:12] <Erik Kline_web_837> DNSSEC?
[12:54:26] <Jen Linkova_web_577> Ole: I don't think I did. Can't call doing this. Was it at whiskey BoF?
[12:54:35] Zhenbin Li_web_374 joins the room
[12:54:36] <Erik Nygren_web_160> Practically should there be a row here for Firewalls?  Ignoring beyond the limit vs dropping isn't viable/good there.
[12:54:45] <Jen Linkova_web_577> Eri{c,k}: ...and RADIUS ;)
[12:54:57] <Jared Mauch_web_312> Ole: I wouldn't be sad if we deprecated HBH, EH, MLD
[12:54:58] <Jen Linkova_web_577> s/call/recall/
[12:55:06] <Jared Mauch_web_312> (and fragments)
[12:55:23] <Ole Trøan_web_853> Jared: The 2022 IPv6 spring clean! ;-)
[12:55:32] <Jen Linkova_web_577> Ole: SPRING???
[12:55:44] <Barbara Stark_web_218> LOL
[12:55:44] Guozhen Dong_web_874 joins the room
[12:55:48] <Jared Mauch_web_312> he means spring cleaning, not SPRING ...
[12:55:58] Martin Vigoureux_web_712 leaves the room
[12:56:11] <Shuping Peng_web_635> this limit may depend on the time scale which we choose to consider
[12:56:12] <Ole Trøan_web_853> I only noticed the potential ambiguity after hitting enter.
[12:56:24] <Darren Dukes_web_339> @bob I said “practical” ;) there has been no practical case, probably not one after 20 years.
[12:56:51] Xueyan Song_web_193 leaves the room
[12:57:00] <Ole Trøan_web_853> Regarding limits. Isn't the limit going to be very use case specific? Do we need a generic one?
[12:57:01] <Jen Linkova_web_577> Darren: I've heard people saying exactly the same about Ipv6 in general..
[12:57:09] Xueyan Song_web_611 joins the room
[12:57:15] <Bob Hinden_web_252> @Darren   Just like Source Routing :-)
[12:57:16] Richard Vallee_web_770 joins the room
[12:57:21] Gabriel Montenegro_web_335 leaves the room
[12:57:59] <Ole Trøan_web_853> Now guys be nice.
[12:58:23] <Erik Nygren_web_160> @Jen: you need HPH PMTUD to avoid the Fragments because probing for HPH options interfered with PMTUD probing?
[12:58:48] Guozhen Dong_web_874 leaves the room
[12:59:12] <Darren Dukes_web_339> @bob Jen LOL :)
[12:59:14] Ana Custura_web_845 joins the room
[12:59:16] <Jen Linkova_web_577> ErikL I'm afraid sometimes fragments are unavoidable
[12:59:23] Xueyan Song_web_611 leaves the room
[12:59:27] Xueyan Song_web_263 joins the room
[12:59:35] Fred Baker_web_995 leaves the room
[12:59:36] Zafar Ali_web_772 leaves the room
[12:59:44] <Jeffrey Haas_web_880> at some point you devolve into a pmtu query to see what a given end to end session MIGHT need to reserve space for in order for a middle box to muck with the packet intrusively with hbh
[13:00:03] Dhruv Dhody_web_231 leaves the room
[13:00:06] Fred Baker_web_572 joins the room
[13:00:07] Dhruv Dhody_web_918 joins the room
[13:00:23] Ran Chen_web_673 leaves the room
[13:00:27] Ran Chen_web_335 joins the room
[13:00:45] Shuai Zhang_web_454 leaves the room
[13:00:48] Jiamin_web_634 leaves the room
[13:00:52] Zhenbin Li_web_374 leaves the room
[13:01:11] Wim Henderickx_web_606 leaves the room
[13:01:21] Ran Chen_web_335 leaves the room
[13:01:25] Jyrki Soini_web_336 joins the room
[13:01:36] Nathalie Trenaman_web_751 leaves the room
[13:01:42] Dawei Fan_web_992 leaves the room
[13:01:56] Doug Montgomery_web_262 joins the room
[13:02:10] Gyan Mishra_web_112 leaves the room
[13:02:23] Xueyan Song_web_263 leaves the room
[13:03:13] <Tim Chown_web_530> maybe bob's audio is in extension headers
[13:03:21] <Tom Herbert_web_345> Ole, there are recommended default for all limits. RFC8504 defines some limits for receiving hosts.
[13:03:22] <mcr> "not working well"... haha. major understatement.
[13:04:23] <Warren Kumari_web_874> @mcr: :-)
[13:04:36] Steven Hartley_web_450 leaves the room
[13:04:40] Steven Hartley_web_656 joins the room
[13:04:42] <Warren Kumari_web_874> CSMA/CD to me always seems like an acronym win.
[13:05:10] <Jen Linkova_web_577> mcr: I'd say '!(working well)' is true in this case, so...
[13:05:10] Steven Hartley_web_656 leaves the room
[13:05:14] Steven Hartley_web_763 joins the room
[13:05:23] <Tom Herbert_web_345> !


Adrian Farrel

Aihua Liu

Alessandro Amirante


Ana Custura

Andrew Alston

Antoine Fressancourt

Balazs Varga

Barbara Stark

Ben Maddison

Bhavit Shah

Bill Fenner

Bob Hinden

Chang Liu

Cheng Li

Chenhao Ma

Chi-Jiun Su

Chongfeng Xie

Daniam Henriques

Daniel Huang

Darren Dukes

David Pelton

Dhruv Dhody

Donald Eastlake

Eduard V

Eric Rosenberg

Éric Vyncke

Erik Kline

Erik Nygren

Fan Yang

Fred Baker

Gang Yan

Geir Egeland

Geraldo Júnior

Giuseppe Fioccola

Gorry Fairhurst

Greg Mirsky

Gunter Van de Velde

Guozhen Dong

Gyan Mishra

Haibo Wang

Haojie Wang

Haoyu Song

Himanshu Shah

Ignas Bagdonas

Jeffrey Haas

Jen Linkova


Jie Dong

Jim Uttaro

Jingrong Xie

Jinyou Dai

Joao Damas

John Border

Jordi Palet Martinez

Juan Cerezo

Jurgen Sonnenmoser

Justin Iurman

Jyrki Soini

Ketan Talaulikar

Kiran Makhijani

Kireeti Kompella

Liu Yao

Loa Andersson

Marco Davids

Mark Andrews

Martin Horneffer

Martin Vigoureux


Michael Richardson

Michael zhou

Mohamed Boucadair

Nicholas Gajcowski

Ole Trøan
[13:05:34] <Tom Herbert_web_345> Sorry!
[13:05:50] <Kireeti Kompella_web_138> that was tom checking pmtu
[13:05:51] <Ole Trøan_web_853> Tom, testing limits in cut and paste? ;-)
[13:05:53] <Jen Linkova_web_577> wow how did you do that??
[13:06:05] <Jared Mauch_web_312> for PMTU issues, i mostly see people have issues with L2 devices than L3 devices
[13:06:11] <Tom Herbert_web_345> Ole, 5am in the morning :-)
[13:06:13] Toerless Eckert_web_214 joins the room
[13:06:24] Steven Hartley_web_763 leaves the room
[13:06:28] Steven Hartley_web_370 joins the room
[13:06:37] Gang Yan_web_110 leaves the room
[13:07:10] <Justin Iurman_web_530> @Tom each in turn :-D
[13:07:16] Mike Koldychev_web_763 joins the room
[13:07:18] Himanshu Shah_web_526 leaves the room
[13:07:22] Himanshu Shah_web_164 joins the room
[13:07:43] Qin Wu_web_863 leaves the room
[13:08:03] Zheng Zhang_web_386 leaves the room
[13:08:08] Steffen Klassert_web_712 leaves the room
[13:08:12] Steffen Klassert_web_452 joins the room
[13:08:53] <Eduard V_web_746> About PMTUD HbH: if it is the closed domain when why not to decrease filtering of ICMP PTB? then normal 8201 would start to operate.
[13:09:16] Steven Hartley_web_370 leaves the room
[13:09:44] Steffen Klassert_web_452 leaves the room
[13:09:48] Steffen Klassert_web_460 joins the room
[13:09:54] <Ole Trøan_web_853> A benefit of PMTU HBH is that it follows the end to end path. No packet from random intermediate source.
[13:10:19] <Justin Iurman_web_530> "header can be large" ... as other options
[13:10:34] Jinyou Dai_web_363 leaves the room
[13:10:36] <Justin Iurman_web_530> "header changes its size at every hop" ... so we're talking about the incremental trace, not the pre-allocated trace
[13:10:38] Jinyou Dai_web_204 joins the room
[13:10:41] <Bob Hinden_web_252> @Ole   +1   Better trust relationship and less likely to be dropped.
[13:11:05] Daniel Huang_web_230 leaves the room
[13:11:08] Zheng Zhang_web_986 joins the room
[13:11:08] <Eduard V_web_746> Ole, ICMP PTB would be generated from the path only. No any urrelevant hops.
[13:11:09] Daniel Huang_web_213 joins the room
[13:11:17] Zheng Zhang_web_986 leaves the room
[13:11:19] <Éric Vyncke_web_747> @Justin changing header size is indeed troubling
[13:11:21] Zheng Zhang_web_797 joins the room
[13:11:59] Zheng Zhang_web_797 leaves the room
[13:11:59] <Jen Linkova_web_577> Also, intermediate routers usually rate limit ICMP generation
[13:12:07] <Éric Vyncke_web_747> @Ole same path if ECMP 'hash' is able to jump over the HbH to do the same 5-tuple hash
[13:12:16] <Gorry Fairhurst_web_638> @Eduard: You could configure a consistent MTU :-); you could enable ICMP everywhere... which might give you a way to reduce the PMTU; To increase you'd need to probe with different size packets - if you had jumboframes that might be less easy to guess what size works.
[13:12:25] Steven Hartley_web_868 joins the room
[13:12:27] Yuehua Wei_web_447 leaves the room
[13:12:29] <Tom Herbert_web_345> @Ole, PMTU probably should be usedon all packets for a flow since the presence of the EH could affect how the packet is routed
[13:13:10] <Jen Linkova_web_577> Eric: we should have start using flow labels finally (it would very, very nice for fragments, esp with anycast ;)
[13:13:25] <Gorry Fairhurst_web_638> @Tom - is the use of alternate routes based on HBH a common problem?
[13:13:36] <Ole Trøan_web_853> @Tom yes, it would have been nice if we could have come up with a mechanism that only required processing if the packet was too large. Large probe with packet truncation was one idea.
[13:13:56] <Éric Vyncke_web_747> @Jen if only... else I agree 100% !
[13:14:25] Justin Iurman_web_530 leaves the room
[13:14:25] <Jen Linkova_web_577> Eric: AFAIK at least some router vendors do that. And my laptop seems to use flow lables
[13:14:32] Justin Iurman_web_876 joins the room
[13:14:49] Jim Uttaro_web_528 leaves the room
[13:15:02] <Éric Vyncke_web_747> @Gorry: not all routers can parse over the HbH to find the L4 header to compute the 5-tuple hash for ECMP
[13:15:13] <Warren Kumari_web_874> Clips for me
[13:15:14] <Barbara Stark_web_218> Definitely clipping
[13:15:34] <Éric Vyncke_web_747> @Jen: indeed, many routers include flow label in their 6-tuple hash
[13:15:35] Steffen Klassert_web_460 leaves the room
[13:15:36] Jie Dong_web_467 leaves the room
[13:15:39] Steffen Klassert_web_759 joins the room
[13:15:40] Jie Dong_web_391 joins the room
[13:15:49] Cheng Li_web_979 leaves the room
[13:15:49] <Tom Herbert_web_345> @Gorry, the extension header might push the L4 headers to far down in packet, so ECMP might have different behavior.
[13:15:51] <Darren Dukes_web_339> @jen use flow label minus any other L4 info.  Do you think that’s possible?.
[13:15:53] Cheng Li_web_722 joins the room
[13:16:42] Kyle Rose_web_715 joins the room
[13:16:57] <Tom Herbert_web_345> @darren, that's the intent. Flow label is usable even when L4 is unreable (e.g. when it's encrypted)
[13:16:58] Jurgen Sonnenmoser_web_167 leaves the room
[13:16:58] Mohamed Boucadair_web_526 leaves the room
[13:17:00] Colby Barth_web_822 joins the room
[13:17:05] <Jen Linkova_web_577> Darren: I might be missing smth but why not? it flow label is empty - use 5 tuple,  otherwise - use flow label...
[13:17:23] <Gorry Fairhurst_web_638> @Eric, Tom: OK. But then ECMP is normally over paths with the same MTU ... which wouldn't be a real impact.
[13:18:00] <Kyle Rose_web_715> Dr. Pedanticus here. "Clipping" refers to when someone's audio is overdriven to the point where it hits the maximum amplitude supported by the encoding. Also called "brickwalling", it produces something that sounds like static or distortion.
What was happening here is better termed "cutting out".
[13:18:01] <Éric Vyncke_web_747> except in very badly designed networks ;-) Hoping that those are the exceptions
[13:18:13] <Gorry Fairhurst_web_638> My suggeted use is that the info returned  is anyway advice about how to probe to check the path - the communication doesn't break becasue the minPMTU is wrong, but it will probes unusessfully with a probe packet of that size
[13:18:22] <Darren Dukes_web_339> @jen I don’t disagree.
[13:18:24] <Ole Trøan_web_853> @Kyle: Thank you!
[13:18:29] Jie Dong_web_391 leaves the room
[13:18:33] Jie Dong_web_336 joins the room
[13:19:09] Zongpeng Du_web_252 leaves the room
[13:19:25] Juhamatti Kuusisaari_web_389 leaves the room
[13:19:29] Juhamatti Kuusisaari_web_207 joins the room
[13:19:50] Jinyou Dai_web_204 leaves the room
[13:19:54] Jinyou Dai_web_258 joins the room
[13:20:10] Zongpeng Du_web_576 joins the room
[13:20:13] <Tom Herbert_web_345> @Gorry, ECMP is a local decision but downstream paths could have very different characteristics (like differnt RTT causing OOO or maybe even drop). Chanigng routing mid flow can be insidious.
[13:20:23] Daniel Huang_web_213 leaves the room
[13:20:27] Daniel Huang_web_221 joins the room
[13:20:38] Zongpeng Du_web_576 leaves the room
[13:20:42] Zongpeng Du_web_136 joins the room
[13:21:29] <Tim Chown_web_530> As mentioned in 6man, the CERN experiment community considered HBH/DO for marking experiment traffic but with the issues there are instead using flow label.
[13:22:07] Zongpeng Du_web_136 leaves the room
[13:22:11] Zongpeng Du_web_182 joins the room
[13:22:22] <Tom Herbert_web_345> @Tim, is this by repurposing bits in the flow label?
[13:22:26] <Gorry Fairhurst_web_638> @Tom I know. There are more evil things out there - such as inconsistent receive size at the remote. In the ned if you wnat to avoid black-holing, I'd say you have to probe with a sacrifical (passing) packet.
[13:22:45] <Gorry Fairhurst_web_638> /ned/end/
[13:22:50] <Tim Chown_web_530> yes, Tom, 5 entropy bits, 15 marking bits (for experiment and application)
[13:22:52] <Darren Dukes_web_339> @jen A bit more complex to implement the hash generation with a conditional. But It would be interesting to see what impact that recommendation could have.
[13:22:54] Juhamatti Kuusisaari_web_207 leaves the room
[13:22:58] Juhamatti Kuusisaari_web_342 joins the room
[13:23:16] Zongpeng Du_web_182 leaves the room
[13:23:20] Zongpeng Du_web_334 joins the room
[13:23:44] Xueyan Song_web_138 joins the room
[13:23:55] <Tim Chown_web_530> @tom they are also using out of band UDP firefly packets which then works for ipv4, though the majority of CERN experiment traffic is now v6.
[13:24:09] Juhamatti Kuusisaari_web_342 leaves the room
[13:24:13] Juhamatti Kuusisaari_web_485 joins the room
[13:24:15] <Tom Herbert_web_345> @Tim, that's going to be a hard sell in 6man. Flow label is deployed and going to be hard to redefine the bits.
[13:24:16] Zongpeng Du_web_334 leaves the room
[13:24:20] Zongpeng Du_web_511 joins the room
[13:24:44] <Tim Chown_web_530> well, it's working, without 6man approval ;)
[13:25:05] <Tom Herbert_web_345> @Tim, nice :-)
[13:25:06] <Éric Vyncke_web_747> @Tom there is also a stateful mode for flow labels (if I remember correclty) where routers can have state to process the flow label
[13:25:09] <Éric Vyncke_web_747> @Tim :-)
[13:25:38] <Erik Kline_web_837> the docs support a domain doing what it wants with the flow label (in general)
[13:25:44] Ben Maddison_web_733 leaves the room
[13:25:48] Ben Maddison_web_467 joins the room
[13:26:05] <Tim Chown_web_530> the hbh/do issues are documented, so i'll encourage them to write a short info draft, around the problem of marking the traffic for measurement.
[13:26:18] <Bob Hinden_web_252> For encapsulation, I think there is a lot of flexability how the flow label could be used.
[13:26:19] Shunwan Zhuang_web_212 leaves the room
[13:26:50] <Tim Chown_web_530> well, in this case erik the packets flow over much of the world's R&E network infra, so it's a big domain, but generally not commercial internet.
[13:26:52] Boris Khasanov_web_362 joins the room
[13:26:58] Xiao Min_web_866 leaves the room
[13:27:34] <Jeffrey Haas_web_880> Without revealing the parties that are asking, there are many who want to meddle with the flow label.
[13:27:52] <Jared Mauch_web_312> i've watched many of the r&e networks slowly move to NAT-PT and actually do less IPv6 today than they did 5-10 years ago now
[13:28:12] <Jared Mauch_web_312> plus selling their IPv4 space (MIT, Merit are two examples)
[13:28:17] <Tim Chown_web_530> i don't know of any R&E network using nat-pt, but we're very off topic now :)
[13:28:45] <Kireeti Kompella_web_138> the restriction MUST NOT be in # present, but in # required to be processed
[13:28:51] Steffen Klassert_web_759 leaves the room
[13:28:55] Steffen Klassert_web_340 joins the room
[13:29:00] <Tom Herbert_web_345> @Bob, the problem is routers may use the whole flow label for ECMP. If a node is changing some bits then they should expect inconsisten routing (maybe that's okay in some circumstance)
[13:29:01] Ana Custura_web_845 leaves the room
[13:29:24] <Tim Chown_web_530> @tom hence the 5 entropy bits as a compromise
[13:29:59] Zheng Zhang_web_323 joins the room
[13:30:10] Richard Vallee_web_770 leaves the room
[13:30:16] <Ole Trøan_web_853> @Kyle what is the term for the artefact when beginning of the audio stream is cut off? I.e when participants just come off mute.
[13:30:36] <Tim Chown_web_530> clipping?
[13:31:06] <Srihari Sangli_web_758> I agree with the previous comment, unless there is a driving application, the vendors' will not support HBH and do in fast path
[13:31:08] <Éric Vyncke_web_747> @Tim and hopping that your routers use all 20 bits and not only a subset ;-) and the hash works fine for 5 bits ;-)
[13:31:13] <Kyle Rose_web_715> @ole: no idea :-)
[13:31:27] Steffen Klassert_web_340 leaves the room
[13:31:31] Steffen Klassert_web_878 joins the room
[13:31:32] <Kyle Rose_web_715> I think that might be "meetecho delay caused by releasing the microphone"
[13:31:38] <Tim Chown_web_530> @eric i saw a draft saying certin vendor uses 16 bits.  cisco? :)
[13:31:39] Shuping Peng_web_635 leaves the room
[13:31:43] Shuping Peng_web_391 joins the room
[13:31:50] XiPeng Xiao_web_253 joins the room
[13:32:05] <Jared Mauch_web_312> (router vendors or network operator configuration?)
[13:32:08] Mike Koldychev_web_763 leaves the room
[13:32:09] Luis Contreras_web_641 joins the room
[13:32:57] Yasunobu Toyota_web_520 leaves the room
[13:33:01] Yasunobu Toyota_web_580 joins the room
[13:33:14] Behcet Sarikaya_web_237 joins the room
[13:33:21] <Roland Bless_web_876> Rehabilitation seems to be worthwhile. While I understand that defining limits may help acceptance for HbH processing, I agree with Eduard that enforcing limits now may become a problem in the future then.
[13:33:56] <Éric Vyncke_web_747> @Roland: indeed, limits in 2021 are not the same as in 2031
[13:34:04] <mcr> was the question of which WG would do this work discussed/resolved?
[13:34:15] <mcr> I'm in favour of rehabilitation.
[13:34:28] <Erik Kline_web_837> limits defined as a function of the MTU would seem to make sense
[13:34:40] <Kyle Rose_web_715> Rehabbing HbH headers is definitely worth some effort. I wish we had the greasing concept back in the early 2000's when v6 stacks were being implemented....
[13:34:42] <Erik Kline_web_837> larger MTU, larger HbH limits
[13:34:44] <Ole Trøan_web_853> @mcr: Currently there are two drafts in queue to be adopted in 6man and one draft adopted in v6ops.
[13:34:53] BEHCET SARIKAYA joins the room
[13:35:08] <mcr> I'd almost be in favour of a new WG.
[13:35:20] <Erik Kline_web_837> 6hbh
[13:35:37] <Ole Trøan_web_853> Hop Scotch WG?
[13:35:42] <Jen Linkova_web_577> Hey WG ;)
[13:35:47] <Kireeti Kompella_web_138> 6ext?
[13:35:53] <Tim Chown_web_530> hoppy eyeballs?
[13:35:56] <mcr> 666
[13:36:06] Zafar Ali_web_842 joins the room
[13:36:08] <Erik Kline_web_837> Tim wins, I think
[13:36:15] <Tom Herbert_web_345> @Eric, we can always revist limits. On the other hand, we do have forty years experience with Internet, we can make some reasonable extrapolations on evolution-- for instance the rate at which new options are being defined for various protocols.
[13:36:27] <Fred Baker_web_572> If I am following the discussion at all, the answer to the first question is "yes"
[13:36:44] <Éric Vyncke_web_747> @Tom agreed but not very flexible
[13:37:18] <Tom Herbert_web_345> @Tom too much fllexibility is the problem :-)
[13:37:22] <Erik Kline_web_837> if we define the limits as a function of the MTU then the limits can adjust as the space to hold them adjusts
[13:37:47] <Éric Vyncke_web_747> @Erik still the number of options could be a problem though
[13:37:56] <Kireeti Kompella_web_138> @fred agreed with the caveat that nbn must not require slow path processing
[13:38:09] Juhamatti Kuusisaari_web_485 leaves the room
[13:38:13] Juhamatti Kuusisaari_web_412 joins the room
[13:38:19] <Fred Baker_web_572> Violent agreement
[13:38:52] <Fan Yang_web_793> I was trying to present another IPv6 HbH use case where is used for DetNet/Raw applications.
Three options are defined to provide detnet flow identification in the forwarding plane, detnet OAM, and DetNet packet replication elimination ordering functions.
The links are
we expect further discussions on applications and collaborations to move it forward.
[13:38:55] <Kireeti Kompella_web_138> disagree with limits based on mtu.
[13:39:13] Guozhen Dong_web_922 joins the room
[13:39:16] Craig Pearce_web_812 joins the room
[13:39:19] <Tom Herbert_web_345> @Eric, 1 option limit is too restrictive, 100 is too permissive. We need to find x where 1 < x < 100.
[13:39:37] <Éric Vyncke_web_747> @Tom ;-)
[13:39:42] Weiqiang Cheng_web_162 leaves the room
[13:40:01] Juhamatti Kuusisaari_web_412 leaves the room
[13:40:02] <Kireeti Kompella_web_138> @tom MUST process one, MAY process more might work
[13:40:09] Srihari Sangli_web_758 leaves the room
[13:40:13] Srihari Sangli_web_916 joins the room
[13:40:18] Srihari Sangli_web_916 leaves the room
[13:40:20] <Éric Vyncke_web_747> @warren +1
[13:40:22] Srihari Sangli_web_292 joins the room
[13:40:30] Steffen Klassert_web_878 leaves the room
[13:40:34] Steffen Klassert_web_629 joins the room
[13:40:36] Guozhen Dong_web_922 leaves the room
[13:40:40] Guozhen Dong_web_842 joins the room
[13:40:41] Richard Li_web_573 joins the room
[13:41:04] <Justin Iurman_web_876> @Éric Is there any use/business case with multiple Hbh options carried at the same time currently? Given their "limited domain" purposes in general, we could assume that it'd still make sense in 10 years or so to have only 1 option per hbh. Also, Ole's comment on making hbh terminal could be a solution. Again, is there any use/business case out there requiring hbh and route/dst opt? If so, some might encapsulates others (e.g., SRv6 carrying IOAM, etc)
[13:41:07] <Gorry Fairhurst_web_638> @Tom - 1 might be made to work soon, MAY process more might work amongst those with enthusiasm.
[13:41:08] <Kyle Rose_web_715> WG: "hoho"
[13:41:15] <Tom Herbert_web_345> @Kireeti, to the extent that the problem is processing the options, not the very presence of the options (i.e. the headers too big problem)
[13:41:48] <Éric Vyncke_web_747> @Justin, e.g., iOAM & MTU options ?
[13:42:15] <Kireeti Kompella_web_138> @tom isn't the DoS aspect of processing hbh the reason why they are configured to be dropped?
[13:42:21] Rakesh Gandhi_web_702 leaves the room
[13:42:25] Rakesh Gandhi_web_677 joins the room
[13:42:29] Juhamatti Kuusisaari_web_917 joins the room
[13:42:59] Steffen Klassert_web_629 leaves the room
[13:43:03] Steffen Klassert_web_415 joins the room
[13:43:03] Adrian Farrel_web_338 leaves the room
[13:43:15] Srihari Sangli_web_292 leaves the room
[13:43:19] Srihari Sangli_web_620 joins the room
[13:43:28] Guozhen Dong_web_842 leaves the room
[13:43:31] <Kireeti Kompella_web_138> +1 to ron
[13:44:50] <Kireeti Kompella_web_138> @ron deprecate router alert and ALL slow path processing
[13:45:06] Guozhen Dong_web_892 joins the room
[13:45:23] <Justin Iurman_web_876> @Éric What about a restriction (x=1) then? They do not necessarily need to be carried altogether
[13:45:29] James Evans_web_181 leaves the room
[13:45:32] <Tom Herbert_web_345> @Kireeti, as I mentioned there are some routers that require reading L4 headers or they'll drop. There was quite a bit of discussion when we worked in the operational implications on extension headers draft.
[13:45:34] Geng-Da Tsai_web_850 leaves the room
[13:45:53] <Tom Herbert_web_345> @Kireeti, +1 on eliminating slow path
[13:46:04] <Timothy Winters_web_155> If you remove router alert, MLD messages won't be processed by all the current host/routers.
[13:46:10] Yuehua Wei_web_198 joins the room
[13:46:12] <mcr> "fast path" and the "not-so-fast path"... @tom... done.
[13:46:18] <Éric Vyncke_web_747> @tim indeed....
[13:46:25] <Kireeti Kompella_web_138> @tom to clarify, x=1 for hbh, not including L4
[13:46:51] <Timothy Winters_web_155> MLD is going thru an update to aim for internet standard, so maybe we could change this?
[13:46:53] Liu Yao_web_468 leaves the room
[13:46:55] Guozhen Dong_web_892 leaves the room
[13:46:57] Liu Yao_web_528 joins the room
[13:47:04] <Kireeti Kompella_web_138> +1 to tim
[13:47:27] Adrian Farrel_web_976 joins the room
[13:47:34] <Kireeti Kompella_web_138> could -> should -> MUST
[13:47:58] Steffen Klassert_web_415 leaves the room
[13:48:02] Steffen Klassert_web_486 joins the room
[13:48:10] <Gorry Fairhurst_web_638> You might allow RA only on a local link... and then not expect this on Internet Routers
[13:48:13] Geng-Da Tsai_web_250 joins the room
[13:48:20] <Bill Fenner_web_948> "just" change MLD to send to all-routers instead of to the group being joined
[13:48:33] <Shuping Peng_web_391> Agree @Eric Vyncke
[13:48:53] <Ole Trøan_web_853> @Bill: Right, any case where that wouldn't work?
[13:49:00] Linda Dunbar_web_585 joins the room
[13:49:56] Uma Chunduri_web_701 joins the room
[13:50:00] Dawei Fan_web_808 joins the room
[13:50:06] Qin Wu_web_690 joins the room
[13:50:57] <Gorry Fairhurst_web_638> @Bill ... I suspect Ethernet Switches might be more tricky to upgrade than many devices...
[13:50:58] <Bill Fenner_web_948> @Ole: Well, MLD snooping switches would have to understand this change too (snooping switches send copies of MLD to the CPU so that they can inspect it and create a tree at layer 2)
[13:51:23] <Gorry Fairhurst_web_638> @Bill <snap> same comment!
[13:51:27] <Bill Fenner_web_948> The quotes in "just" were ironic
[13:51:32] <Bill Fenner_web_948> @Gorry great minds think alike
[13:51:33] Steffen Klassert_web_486 leaves the room
[13:51:36] Jeffrey Haas_web_880 leaves the room
[13:51:37] Steffen Klassert_web_638 joins the room
[13:51:40] Fan Yang_web_793 leaves the room
[13:51:42] <Ole Trøan_web_853> @Bill: at least from the platform guys I checked with they typically didn't do MLD snooping for IPv6 link-local multicast anyway.
[13:51:57] <Warren Kumari_web_874> Yah. I view this session as a really good example of the IETF doing something well :-)
[13:52:13] <Warren Kumari_web_874> If I were feeling snarky, I'd add "... for a change" :-P
[13:52:18] <Éric Vyncke_web_747> @Warren +1 and useful !
[13:52:23] <Bill Fenner_web_948> @Ole: do we only care about IPv6 link-local multicast?  Then we can eliminate MLD altogether
[13:52:33] <Shuping Peng_web_391> Indeed +1 @Warren
[13:52:49] <Ole Trøan_web_853> @Bill: _that_ I don't know. But at least that's one big constituent.
[13:53:02] Eric Rosenberg_web_646 leaves the room
[13:53:04] <Éric Vyncke_web_747> @Bill AFAIK many routers ignore MLD for link-local multicast
[13:53:18] Tarek Saad_web_667 joins the room
[13:53:18] <Ole Trøan_web_853> @Warren: Now you ruined it by being snarky! ;-)
[13:53:33] Guozhen Dong_web_899 joins the room
[13:54:27] Steffen Klassert_web_638 leaves the room
[13:54:31] Steffen Klassert_web_633 joins the room
[13:54:42] Ketan Talaulikar_web_611 leaves the room
[13:55:02] <Shuping Peng_web_391> yes, the draft could start with a good analysis and research first
[13:55:09] Kiran Makhijani_web_814 leaves the room
[13:55:13] <Shuping Peng_web_391> the deprecated one
[13:55:22] <Boris Khasanov_web_362> +1 to make such draft with analysis
[13:55:24] <Bill Fenner_web_948> @Ole, @Éric: in 1999 the design was specifically that link-local was required for MLD, but, that thought was not forceful enough to make it into routers
[13:55:28] <Kireeti Kompella_web_138> can some providers write down WHY they drop hbh on ingress (if indeed they do)
[13:55:41] <Darren Dukes_web_339> I think the trouble hbh has is that options only have significant use in limited domains. Internet wide it is not likely to be widely used and this has been observed, which makes meetings like this seem necessary. I wonder if we can make that distinction in any drafts that move forward.
[13:56:03] <Ole Trøan_web_853> @Bill: I don't think MLD makes any sense for routers for link-local scoped. But it does for L2 switches
[13:56:09] Ville Hallivuori_web_531 leaves the room
[13:56:09] <Eduard V_web_746> Kireeti, they do not want to test and support. Minimize OPEX and CAPEX.
[13:56:13] <Ole Trøan_web_853> Which I don't think couldn't make it scale....
[13:56:13] Steffen Klassert_web_633 leaves the room
[13:56:16] <Shuping Peng_web_391> Thank you, all!
[13:56:17] Steffen Klassert_web_972 joins the room
[13:56:22] Rob Hamilton_web_939 leaves the room
[13:56:22] Loa Andersson_web_674 leaves the room
[13:56:22] Kyle Rose_web_715 leaves the room
[13:56:22] Balazs Varga_web_308 leaves the room
[13:56:24] Colby Barth_web_822 leaves the room
[13:56:24] Steven Hartley_web_868 leaves the room
[13:56:24] Daniam Henriques_web_630 leaves the room
[13:56:24] <Boris Khasanov_web_362> Thanks and bye!
[13:56:24] Juhamatti Kuusisaari_web_917 leaves the room
[13:56:24] <Warren Kumari_web_874> Please note: if you use the term "limited domain" it needs to be **very** clearly defined.
[13:56:25] Michael Richardson_web_400 leaves the room
[13:56:27] Ben Maddison_web_467 leaves the room
[13:56:27] Chongfeng Xie_web_350 leaves the room
[13:56:28] Geir Egeland_web_487 leaves the room
[13:56:28] Jyrki Soini_web_336 leaves the room
[13:56:29] Chi-Jiun Su_web_266 leaves the room
[13:56:29] <Erik Kline_web_837> thanks
[13:56:29] <Kireeti Kompella_web_138> thanks, chairs!
[13:56:30] Greg Mirsky_web_771 leaves the room
[13:56:30] Andrew Alston_web_231 leaves the room
[13:56:32] Daniel Huang_web_221 leaves the room
[13:56:33] Eduard V_web_746 leaves the room
[13:56:34] Ole Trøan_web_853 leaves the room
[13:56:34] <Bill Fenner_web_948> @Ole: The design was specifically that hosts were required to send MLD for link-local, for the benefit of switches.
[13:56:34] <Timothy Winters_web_155> thanks!
[13:56:34] Ron Bonica_web_321 leaves the room
[13:56:36] <Éric Vyncke_web_747> thanks
[13:56:36] <Yuya Kawakami_web_504> Thanks all and chairs
[13:56:36] Dhruv Dhody_web_918 leaves the room
[13:56:39] Yisong Liu_web_130 leaves the room
[13:56:39] Richard Li_web_573 leaves the room
[13:56:39] Roland Bless_web_876 leaves the room
[13:56:40] Aihua Liu_web_359 leaves the room
[13:56:40] <Tom Herbert_web_345> Thanks chairs for a great session!
[13:56:41] <Warren Kumari_web_874> and what the border is, and how it is enforced, and....
[13:56:41] <Darren Dukes_web_339> @warren ack
[13:56:41] Srihari Sangli_web_620 leaves the room
[13:56:41] <Jen Linkova_web_577> Kireeti: for example, because nobody told them that new version of their software *can* skip EH and look at transport protocol
[13:56:41] <Antoine Fressancourt_web_885> Thanks
[13:56:42] Bob Hinden_web_252 leaves the room
[13:56:42] Yuji Tochio_web_561 leaves the room
[13:56:43] Xueyan Song_web_138 leaves the room
[13:56:45] Yuehua Wei_web_198 leaves the room
[13:56:45] Tomohiro Fujisaki_web_149 leaves the room
[13:56:46] Paolo Volpato_web_834 leaves the room
[13:56:47] Éric Vyncke_web_747 leaves the room
[13:56:50] Erik Kline_web_837 leaves the room
[13:56:51] Warren Kumari_web_874 leaves the room
[13:56:53] Adrian Farrel_web_976 leaves the room
[13:56:53] Haoyu Song_web_436 leaves the room
[13:56:55] Behcet Sarikaya_web_237 leaves the room
[13:56:59] Doug Montgomery_web_262 leaves the room
[13:56:59] Timothy Carlin_web_359 leaves the room
[13:56:59] Rurick Kellermann_web_449 leaves the room
[13:57:00] Stefano Previdi_web_656 leaves the room
[13:57:00] Jinyou Dai_web_258 leaves the room
[13:57:02] Radu Valceanu_web_642 leaves the room
[13:57:02] Craig Pearce_web_812 leaves the room
[13:57:02] Tarek Saad_web_667 leaves the room
[13:57:02] Marco Davids_web_406 leaves the room
[13:57:03] Pascal Thubert_web_135 leaves the room
[13:57:03] Gorry Fairhurst_web_638 leaves the room
[13:57:03] Alessandro Amirante_web_992 leaves the room
[13:57:03] abdullahalshoaili_web_636 leaves the room
[13:57:03] Giuseppe Fioccola_web_457 leaves the room
[13:57:03] Haibo Wang_web_335 leaves the room
[13:57:03] Bill Fenner_web_948 leaves the room
[13:57:03] Jen Linkova_web_577 leaves the room
[13:57:03] Chang Liu_web_210 leaves the room
[13:57:03] Barbara Stark_web_218 leaves the room
[13:57:03] xiebin_web_676 leaves the room
[13:57:03] John Border_web_604 leaves the room
[13:57:03] Tim Chown_web_530 leaves the room
[13:57:03] Martin Horneffer_web_622 leaves the room
[13:57:03] Tianji Jiang_web_191 leaves the room
[13:57:03] Bhavit Shah_web_106 leaves the room
[13:57:03] Chenhao Ma_web_963 leaves the room
[13:57:03] Donald Eastlake_web_579 leaves the room
[13:57:03] Yoshifumi Atarashi_web_118 leaves the room
[13:57:03] Tom Hill_web_113 leaves the room
[13:57:03] Tom Herbert_web_345 leaves the room
[13:57:03] Torunn Narvestad_web_449 leaves the room
[13:57:03] Gunter Van de Velde_web_638 leaves the room
[13:57:03] Stuart Cheshire_web_989 leaves the room
[13:57:03] Michael zhou_web_534 leaves the room
[13:57:03] David Pelton_web_577 leaves the room
[13:57:03] Yuya Kawakami_web_504 leaves the room
[13:57:03] Kireeti Kompella_web_138 leaves the room
[13:57:03] Antoine Fressancourt_web_885 leaves the room
[13:57:03] Haojie Wang_web_203 leaves the room
[13:57:03] mackermann@bcbsm.com_web_126 leaves the room
[13:57:03] Pablo Camarillo_web_900 leaves the room
[13:57:03] Ignas Bagdonas_web_490 leaves the room
[13:57:03] Timothy Winters_web_155 leaves the room
[13:57:03] Mark Andrews_web_969 leaves the room
[13:57:03] Fred Baker_web_572 leaves the room
[13:57:03] Paul Bottorff_web_796 leaves the room
[13:57:03] Jared Mauch_web_312 leaves the room
[13:57:03] Jie Dong_web_336 leaves the room
[13:57:03] Jordi Palet Martinez_web_514 leaves the room
[13:57:03] Geraldo Júnior_web_482 leaves the room
[13:57:03] Boris Khasanov_web_362 leaves the room
[13:57:03] Himanshu Shah_web_164 leaves the room
[13:57:03] Juan Cerezo_web_708 leaves the room
[13:57:03] Xia Chen_web_701 leaves the room
[13:57:03] Ahmed Bashandy_web_107 leaves the room
[13:57:03] Zongpeng Du_web_511 leaves the room
[13:57:03] Toerless Eckert_web_214 leaves the room
[13:57:03] Yasunobu Toyota_web_580 leaves the room
[13:57:03] chenjinqiang_web_671 leaves the room
[13:57:03] Zafar Ali_web_842 leaves the room
[13:57:03] Rakesh Gandhi_web_677 leaves the room
[13:57:03] Liu Yao_web_528 leaves the room
[13:57:03] Ching-Heng Ku_web_671 leaves the room
[13:57:03] Joao Damas_web_350 leaves the room
[13:57:03] Shuping Peng_web_391 leaves the room
[13:57:03] Geng-Da Tsai_web_250 leaves the room
[13:57:03] Uma Chunduri_web_701 leaves the room
[13:57:03] Vishnu Beeram_web_373 leaves the room
[13:57:03] Luis Contreras_web_641 leaves the room
[13:57:03] Quan Xiong_web_238 leaves the room
[13:57:03] Linda Dunbar_web_585 leaves the room
[13:57:03] Qin Wu_web_690 leaves the room
[13:57:03] XiPeng Xiao_web_253 leaves the room
[13:57:03] Justin Iurman_web_876 leaves the room
[13:57:03] Michael Breuer_web_400 leaves the room
[13:57:03] Sheng Liu_web_683 leaves the room
[13:57:03] Guozhen Dong_web_899 leaves the room
[13:57:03] Erik Nygren_web_160 leaves the room
[13:57:03] Cheng Li_web_722 leaves the room
[13:57:03] Steffen Klassert_web_972 leaves the room
[13:57:03] Stewart Bryant_web_669 leaves the room
[13:57:03] Darren Dukes_web_339 leaves the room
[13:57:03] Dawei Fan_web_808 leaves the room
[13:57:03] Zheng Zhang_web_323 leaves the room
[13:58:47] leaves the room
[14:01:01] Meetecho leaves the room
[14:06:39] alexamirante leaves the room
[14:19:04] BEHCET SARIKAYA leaves the room
[14:23:56] mcr leaves the room