wait a minute here

Thu Sep 5 21:54:45 UTC 1996

> 
> > 
> > Someone just wrote me and said:
> > 
> > > I would be happy to remove a lot of out /24's if the NIC would allocate a
> > > larger than /32 to renumber into. The NIC does not thing this is as
> > > important.
> > 
> > According to what I know of InterNIC's policies, they will be glad to trade
> > aggregated address space for unaggregated space, _in_roughly_equal_parts_,
> > and assuming that you have efficiently used your old space.
> > 
> > It's not that InterNIC doesn't consider this important -- rather than they
> > are not in the routing business and the routability of addresses is not one
> > of the criteria they can look at when making allocations.
> 
> I don't think this is particularly consistent with what the NIC folks said
> at the last NANOG. I thought they said that they turn away any request for
> less than a /19 (32 class Cs) and that although they can't legally justify 
> this it is not hard to enforce because it is consistent with Sprint's filters.

What we said is that we refer all those requests for less than a /19 to
their ISP.  Once we explain in detail the routing and filtering issues,
most everyone agrees.  

> > 
> > So don't put it to them in terms of routability, just SWIP your suballocat-
> > ions and write a nice polite letter showing how some parts of the world will
> > be better off and no part of the world will be worse off if prefixes X and Y
> > are returned to the pool in exchange for prefix Z (of size ~X + ~Y).
> 
> Even though routabilty is not the NIC's problem, they have used routability
> to define policy.

What we've tried to do is help educate the public.  The routability issues
exist and the InterNIC cannot put their head in the sand and pretend they
don't, however, neither will we totally ignore the problem of the IPv4 space
being a limited resource.  We need to be *and are* cognizant of both.  

Kim

> 
> Jim
> 