> What I am very concerned about is that we are focused on the wrong problem. > We don't have a problem flooding a large number of LSPs on a single interface.
The fact that there is undocumented knowledge that completely contradicts established standards that is necessary for scalable operations implies that there is indeed a problem. GV> I agree with Tony. Maybe the ISISoTCP draft is addressing the wrong problem, but the WG should either explain or should not use undocumented knowledge as argument. Undocumented is undocumented. Undocumented information is not a normative position to say "only the happy few know, and we are not explaining". That seems not a nice and friendly and appropriate position to defend. The authors of "draft-hsmit-lsr-isis-flooding-over-tcp-00" tried to deliver an educated objective description and explanation why there is an issue with the current standard behavior, and sofar I did not receive any indication at all, what is wrong or incorrect with the documented issues in the draft-hsmit-lsr-isis-flooding-over-tcp-00 draft. G/ -----Original Message----- From: Lsr <lsr-boun...@ietf.org> On Behalf Of tony...@tony.li Sent: Thursday, November 8, 2018 04:40 To: Les Ginsberg <ginsb...@cisco.com> Cc: lsr@ietf.org; Henk Smit <hhws...@xs4all.nl> Subject: Re: [Lsr] Teasing us with secrets Les, > I am not "teasing". > I also don't think this is much of a secret. > > I am pretty sure experienced implementers know what I am talking about. > (BTW, I include you in that group. :-) ) Ongoing undocumented knowledge is wholly irrelevant when writing standards. There’s simply nothing to reference and thus a tease. It’s true that there is no obligation to document this knowledge, but at the same time, that knowledge cannot be used to pre-empt alternative solutions. > What I am very concerned about is that we are focused on the wrong problem. > We don't have a problem flooding a large number of LSPs on a single interface. The fact that there is undocumented knowledge that completely contradicts established standards that is necessary for scalable operations implies that there is indeed a problem. Tony _______________________________________________ Lsr mailing list Lsr@ietf.org https://www.ietf.org/mailman/listinfo/lsr _______________________________________________ Lsr mailing list Lsr@ietf.org https://www.ietf.org/mailman/listinfo/lsr
