[c-nsp] BGP table growth and memory

[Mark Kent]

> I'm surely not the first person to notice this, so I have to wonder what
> kind of plans/forecast other small/medium providers have, other than a
> forklift upgrade.

If you are roughly an EndUser(*) then you can just scale back
by filtering your peers like with prefix-lists that limit
the size of netblocks you want in particular ranges.

For example, list twentythree below lets in less than 114k paths.
List twentytwo lets in 106k paths.

It makes your router look years younger :-)

(*) By "roughly an EndUser" I mean either

a) the end of the line for bgp, and you can point default routes
or
b) if not the end of the line for bgp, then you have like-minded
   and cooperative downstreams.

-mark

<snip out bogon denies>
ip prefix-list twentytwo seq 130 permit 192.0.0.0/5 le 24
ip prefix-list twentytwo seq 140 permit 200.0.0.0/6 le 24
ip prefix-list twentytwo seq 150 permit 204.0.0.0/8 le 24
ip prefix-list twentytwo seq 160 permit 0.0.0.0/0 le 22


<snip out bogon denies>
ip prefix-list twentythree seq 130 permit 192.0.0.0/5 le 24
ip prefix-list twentythree seq 140 permit 200.0.0.0/6 le 24
ip prefix-list twentythree seq 150 permit 204.0.0.0/8 le 24
ip prefix-list twentythree seq 160 permit 0.0.0.0/0 le 23