[ppml] 2005-1 or its logical successor

Paul Vixie paul at vix.com
Fri Oct 28 02:48:04 EDT 2005


# I think it's a mistake to focus on renumbering as the main issue; it really
# comes down to multihoming.  Whether I multihome using PA or PI addresses, my
# prefix is necessarily going to take up a slot in the global routing table.
# So if multihoming is to be allowed at all, routing tables will have to grow.

there are three similar sounding reasons for preferring PI space:

1. multihoming
2. anycasting
3. independence

of these, the first two are indistinguishable by external parties.  i said as
much to david williamson after his "allow a /24 for anycast" policy proposal
got so few votes.  my recommendation there is, craft a policy that treats
anycast and multihoming the same, and it would probably get more traction.

(by "indistinguishable" i mean that from the point of view of the rest of the
internet, there is no difference from you announcing a prefix from many places
because you have your own backbone, or because you're able to generate the
same services from lots of pops using replicated server resources.)

((and note that multihoming can be one-site many-exits, or many-sites
many-exits, the commonality is many-exits, which can be transits or peers or a
mixture.  there are more similarities than differences in this taxonomy.))

"independence" is the thing that specifically covers "fear of renumbering"
and while it may not seem like a large problem to a business of ~100 employees
mostly using DHCP, it's a VERY big deal to a business of ~10000 employees, or
to a tier-2 ISP wishing to sell services to such businesses.  we've heard from
members and others who have said "we don't want our address space to lock us
into an ISP" and i think we have to take that seriously even if some of us
might not share that concern in our own networks.



More information about the ARIN-PPML mailing list