Re: RRG thoughts (was [Bier] [pim] Q on the congestion awareness of routing protocols)

[Robert Raszuk <[e-mail]>]

All,

>  (I have observed BGP RIBs of 60M routes)

This is a classic example of routes being confused with bgp paths.

Jeff - do you have an idea how we can fix this massive confusion industry
wide ?

60M paths you quote above is only an issue in very poor implementations.
Efficient implementations share b_path attributes among b_nets.

Also note that in any decent implementation of ADD-PATHs you can specify
how many paths you want to advertise each with a different next hop - what
is all that really matters in practice.

Then if your peers do not talk ADD-PATHs you can use rfc6774 especially now
when RRs are moving to be virtual running on x86 computes and may have the
ability to be enabled with ORR rfc9107.

Thx,
Robert


On Sat, Dec 17, 2022 at 11:40 PM Jeff Tantsura <[e-mail]>
wrote:

> Speaking as RTGWG chair - we are to continue bringing up
> “interesting/routing relevant” topics to wg meetings. We would also be
> happy to have a more structured (through interims/IETF meetings) dedicated
> time to report/discuss top ideas that came up from the yet to be
> established RRG/SiG/any other form.
> ——————————-
> BGP/TCP part
>
> BGP with Add-Path  ALL used indiscriminately in large networks is indeed a
> nightmare (I have observed BGP RIBs of 60M routes), you can do the math.
> While with some creativity, can be reduced (e.g policy based Add-Path with
> BW aware advertising), deployment of such technologies and relevant tooling
> is anything but trivial.
>
> Cheers,
> Jeff
>
> > On Dec 11, 2022, at 10:08, Adrian Farrel <[e-mail]> wrote:
> > I find myself in agreement with this thread.
> >
> > A few of us (daniel King, Dirk Trossen, me) have been trying to run
> "side meetings" either at IETF meetings (virtual and physical) or as
> interims (virtual) to bring "interesting, routing-related research papers"
> to the IETF community. These meetings have been pretty well attended
> (although IETF-115 was a challenge for scheduling).
> >
> > If I have it right, the Internet routes around problems. I think we can
> do that quite well by self-organising.
> > - This mailing list is a good place for routing discussions (thanks to
> the ADs for making that clear).
> > - We can continue to bring routing research to the IETF community. Let's
> find recent papers and active research and bring the authors to IETF
> meetings or to interims to kick up some discussion.
> > - I know that the RTGWG chairs are happy to have occasional
> presentations in their working group, and even focussed interims if a topic
> is bubbling.
> > - The routing ADs seem to like the idea of including strategic and
> recognised papers in the Area meeting at IETF meetings.
> >
> > I'd be happy to work with others (Luigi, Dirk, Dan, ...) to keep this
> seam of side meetings going.
> >
> > Let's collect three or four proposals for presentations in Yokohama.
> >
> > Best,
> > Adrian
> >
> > -----Original Message-----
> > From: routing-discussion <[e-mail]> On
> Behalf Of Luigi Iannone
> > Sent: 11 December 2022 13:10
> > To: Robert Raszuk <[e-mail]>
> > Cc: Jon Crowcroft <[e-mail]>; Dino Farinacci <
> [e-mail]>; [e-mail]
> > Subject: Re: RRG thoughts (was [Bier] [pim] Q on the congestion
> awareness of routing protocols)
> >
> > Hi,
> >
> >> On 11 Dec 2022, at 00:56, Robert Raszuk <[e-mail]> wrote:
> >>
> >> Joel at all,
> >>
> >>> The idea of a suitable forum for informal conversation between
> >>> researchers, vendors, and operators, seems useful.
> >>
> >> The crux of the matter is that researchers stay around SIGCOMM, vendors
> stick to IETF to push RFCs and real operators prefer to go to local NOGs or
> Apricot/RIPE/NANOG or IXP centric meetings. Then we have a zoo of those
> "OPEN" everything venues.
> >
> > Indeed, the crucial part is the role of the “leaders”, let’s say 2 or 3
> chairs (to use IETF terminology) that spend time looking around and trying
> to bring together these different communities on topics that span among all
> of them. Building an agenda, but also gathering the topics of interest.
> >
> > Another point that I made in the draft I have mentioned is about
> increasing ties with academics from the different venues.
> > IETF 115 was in London, Jon Crowcroft showed up to the side meeting,
> which was good.
> > IETF 116 is in Japan, there are a few researcher there that is worth
> inviting.
> > And so on and so forth for the different places where the IETF  will be
> held.
> > Even in returning places would be nice to have updates “what did happen
> to that nice idea you presented last time…”
> >
> > There is potential for a lot of things….
> >
> > I do believe this is possible and would love to have help to making this
> happen.
> >
> > Ciao
> >
> > L.
> >
> >
> >
> >>
> >> As an industry we have an issue - we no longer communicate on the tech
> level unless you have folks attending and participating in all of the
> above. And I am not guessing here ... I am (trying) to be part of all of
> the above. It is no longer how IETF started in the early days.
> >>
> >> Just take a look at SCION enthusiasts. Observe their struggle to
> educate/sell the idea. I am not yet convinced that what they cooked is
> solid, but the effort they are taking gives you a good view on the present
> situation.
> >>
> >> Cheers,
> >> R.
> >> _______________________________________________
> >> routing-discussion mailing list
> >> [e-mail]
> >> https://www.ietf.org/mailman/listinfo/routing-discussion
> >
> > _______________________________________________
> > routing-discussion mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/routing-discussion
> >
> > _______________________________________________
> > routing-discussion mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/routing-discussion
>