Competition for address allocation
Scott Huddle wrote:
> Why not introduce this same sort of competition to address
> allocation services? I can forsee that you have the sort
> of problem of "Mom said no, so ask Dad", but this exists
> for gTLDs as well. Registries would compete on service
> and price but would have to be "blessed" or "licensed"
> to be in the business. This would also eliminate the
> concerns over pricing as well, the market would set the
> price rather than beaurocrats.
> At 03:19 PM 2/5/97 -0800, Paul A Vixie wrote on NANOG
> >> If you need to know that there will be a registry
> >> for each new gTLD, then I'm willing to run any of
> >> them that cannot otherwise find a home.
> >The new gTLD's are all ultimately going to be shared. There's going to be
> >a crossbar effect where registries and gTLDs are not mapped one to one; any
> >registry will be able, if willing, to register names in any of the new gTLDs.
Yeah. This was what I meant a few days ago when I suggested that perhaps
the SOA (Source of Authority) of the address allocation become part of
the actual IP. I recognize that, in a way, the /8 end of the IP address
(or the /16, or the /19) is actually being treated that way now (the
authority being the hierarchical level from which PI space begins), but
the discussions re: multihoming and address holes lead me to ask
whether, as Scott asks, can this be externalized (without incurring an
unpayable route lookup bandwidth penalty) in practice? (Again, I'm not
sure if this is on-topic vis a vis ARIN). The externalization of IP
source of authority would, it seems to me, render the anticompetitive
fears expressed here (on this list) somewhat less threatening. And I
understand less about the flexibility of the installed base of routers
and their abilities to deal with this than I ought to (admittedly, I
know *nothing* about them operationally!).