Re: [Bier] [pim] Q on the congestion awareness of routing protocols

[Robert Raszuk <[e-mail]>]

HI,

> Are you saying that each router receive 999 copies of
> full route (these days, there are 1M routes) information
> for iBGP is not a problem?

If reflectors are configured with ADD-PATHs ALL - yes this is the case.

However those are not *copies* per se. BGP routes consists of two
main parts: net+paths

So each implementation shares nets across all received routes. And when
paths are identical they are also shared.

Some implementations do it better then others (in terms of paths sharing
efficiency), but this is not really 999 times more RAM.

> OTOH, if there are 1000 border routers, configuration
> simplicity means to have some automatic mechanism to
> generate and install configuration files for the routers,
> with which TCP mesh can be maintained automatically
> without any added complexity.

The issue is that while you are absolutely correct in respect to automated
configuration push when you add or delete an ASBR you do not usually want
to push config to all remaining 998 border routers to add or delete a TCP
session
to a new/old node.

Thx,
R.



On Wed, Dec 7, 2022 at 7:08 AM Masataka Ohta <
[e-mail]> wrote:

> Robert Raszuk wrote:
>
> >> As such, if iBGP by TCP mesh without reflectors is not acceptable,
>
> > See RRs were deployed to address three points:
> > #1 - Configuration simplicity
> > #2 - Path reduction
> > #3 - TCP session reduction
>
>  > In the old days keeping 1000 of TCP connections in custom
>  > kernels were an issue.
>
> Are you saying that each router receive 999 copies of
> full route (these days, there are 1M routes) information
> for iBGP is not a problem?
>
> OTOH, if there are 1000 border routers, configuration
> simplicity means to have some automatic mechanism to
> generate and install configuration files for the routers,
> with which TCP mesh can be maintained automatically
> without any added complexity.
>
>                                         Masataka Ohta
>
> _______________________________________________
> routing-discussion mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/routing-discussion
>