[arin-ppml] simple question about money
jmaimon at chl.com
Tue Jun 10 23:29:33 EDT 2008
Randy Bush wrote:
> end sites are being given /48s etc by their upstreams. those which
> multi-home will announce the /48s to both sides.
Of which there are only 65k in a /32, which means all but the smallest
of ISPs will need more over their effective lifetime, but thats another
> and, when ipv6 becomes
> needed for real business, folk from the business side of the company
> will come walking over and telling us that customers are complaining
> that they can not get their mtv from those /48s. and the filters will
> fall, just as they did with ipv4. and the v4 table is over half /24s.
> this is just business, not an ideal world.
In ideal world, the resources would be appropriately charged for at the
point of consumption. The resources are global routing slots and
registry operational expenses. That routing people rely on the registry
fee to protect access to the routing slots resource is a hack.
However, having the local ISP charge per advertised prefix equates to
local actors selling global slots that weren't theirs to local customers.
Basically the only intelligent course of action for any local actor
would be to sell as many slots as possible, since if they wont, someone
else will - and they will consume the same number of slots on everyone
keeping a DFZ table either way.
Perhaps a centralized enough (PKI or otherwise) routing registry could
actually be viewed as the leasing of a routing slot, assuming sufficient
take-up of filtering based on this registry.
> and, in the meantime, we the incumbents use routing table aggregation as
> a excuse for creating a barrier to entry.
Not sure I understood you correctly. Did you mean that incumbents
perform routing table aggregation to create an entry barrier for small
sized prefixes or did you mean that incumbents have created a PI entry
barrier in response to routing table de-aggregation (or both)?
More information about the ARIN-PPML