any dangers of filtering every /24 on full internet table to preserve FIB space ?

Mon Oct 17 17:02:27 UTC 2022

I already had this idea, I even implemented it in the desperate time of the
512K "bug".
And with that I can tell you:
Do not do it! You will be bothered!

But if you want to go this way, what I can recommend is to try not to put
routes in the FIB that match your Default.

Talking about having a default route other than /dev/null is already a
problem at first...
Because a Transit Provider is not expected to use the Default route.
But I'm not even going to get into that (many flames will arise).

If you really decide to use a default route and choose what will not and
what will not apply to the FIB, you must be prepared for a certain
complexity in these choices.
And the more Peers and DFZ views you have, the more complex it will be.

In a very simplified hypothetical example of dual-homed DFZ, take the best
routes from link B, and leave the default by link A.

There are even tools that, comparing flow analysis and routes exported from
the RIB, "choose" the routes with more matching packages, and apply that
for you.
But thinking in this way, the transition to the dark side of the force is
already beginning to be made. Walking through the valley of death until
arriving in the land of the Route Optimizers.

My memory is not helping me...
But I think the name of one of the projects that did this magic was rt-flow
or flow-rt. Something like.

Em seg., 10 de out. de 2022 às 12:01, Edvinas Kairys <
edvinas.email at gmail.com> escreveu:

> Hello,
>
> We're considering to buy some Cisco boxes - NCS-55A1-24H. That box has
> 24x100G, but only 2.2mln route (FIB) memory entries. In a near future it
> will be not enough - so we're thinking to deny all /24s to save the memory.
> What do you think about that approach - I know it could provide some
> misbehavior. But theoretically every filtered /24 could be routed via
> smaller prefix /23 /22 /21 or etc. But of course it could be a situation
> when denied /24 will not be covered by any smaller prefix.
>
> What do you think about this approach ?
>
> Also maybe you know - some advices for edge routers that have at least
> 8x100G interfaces and "good" memory for prefix count ? Thanks
>

-- 
Douglas Fernando Fischer
Engº de Controle e Automação
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20221017/12c24a59/attachment.html>