Scaling ARIN proposal for small ISPs - and economic reality

[Howard C. Berkowitz]

At 4:31 PM -0800 1/22/97, Kent Crispin wrote:
>Howard C. Berkowitz allegedly said:
>>
>[...]
>>
>> Please give us implementable engineering suggestions about how to avoid
>> buying such a resource from competitors in the next 6-18 months.
>
>I must have missed the part about ARIN going away in 6-18 months...
>
>--

No, that's not what I was saying. I wasn't speaking about ARIN.  I was
responding to complaints that people "should not" have to get address space
from hierarchically higher-level aggregators with larger blocks, because
unnamed routing technology  "ought" to be immune to the factors motivating
provider-based aggregation.

There's certainly a lot of research going on in new routing paradigms.  But
I was referring to the reality that we have no proven alternatives in the
short term.  If there were not a need for aggregation to reduce the routing
table sizes, much of the RFC2050-style incentives for registries to dole
out only large blocks would be inappropriate.

What I was saying was that in the near term, we don't have alternatives,
and we would need a North American registry for the near term.  At such
time as alternative routing products are available that can deal with
infinitely large tables, the whole aggregation paradigm needs to be
revisited, and the guidance for registries will need to change.