BGP dampening parameters
danny at qwest.net
Sat Dec 11 04:48:16 UTC 1999
Do many folks actually implement prefix-based dampening parameters? On
customers or peers only? Seems like more gunk in the router configurations to
confuse folks, although I suppose that it's potentially less confusing than
having customers request that a given prefix be "unsuppressed".
And, _if it wasn't obvious already, for all the folks suggesting that a
globally routable /n (where n = some prefix length) was a requirement, you
should have a look at section 1.5 "Aggregation versus damping". Even if
you're multi-homed and your providers aren't performing any type of specifics
suppression proxy-aggregation type stuff, an aggregate announcement from one
of your upstreams *not* being suppressed could save some headaches.
I don't completely agree with the "bgp fast-external-fallover" section (2.4),
though I suppose it's up to the operator anyways :-)
One useful addition to the document would be regarding BGP Route Refresh
versus the soft-reconfiguration stuff.
Along these lines, is asking why NANOG doesn't publish these types of BCP
documents a silly thing? Given, the BCP from that perspective seems to
suggest publishing an ID and submitting to some random IETF WG, but
> "Other" Sean:
> | Have people adjusted their dampening parameters or are they still using the
> | Cisco defaults?
> Please see http://www.ripe.net/docs/ripe-178.html
More information about the NANOG