ARIN Proposal

[Howard C. Berkowitz]

At 9:42 AM -0800 1/21/97, Kent Crispin wrote:
>David R. Conrad allegedly said:
>>
>> Karl,
>>
>> >Therefore, every ISP must be an ARIN "associate" if they have an ounce of
>> >sense, and they must be able to get those magic /19s (or larger if they can
>> >justify them).
>> >
>> >To fail to provide that on a *level* playing field is going to invite
>> >lawsuits
>>
>> Currently within APNIC and RIPE-NCC, if an organization pays the
>> membersip fees, we allocate (or reserve for) them a /19 block.  While
>> at APNIC (and I assume RIPE) we do try to discourage everyone (not
>> just end users) from getting provider independent blocks from the
>> registry (we have a form letter that says "routers are falling over,
>> blah blah blah"), we will do so if they insist (and they pay the
>> membership fee).
>>
>> In both our cases (not wanting to speak for RIPE-NCC, but I believe this
>> to be the case -- I'm sure they'll blast me if I'm off base), the fees have
>> (apparently) had the effect of discouraging smaller ISPs from obtaining
>> blocks from the registries directly.
>>
>> Do you consider this a level playing field?
>>
>> Regards,
>> -drc
>
>Clearly it is not a level playing field -- it is an unlevel playing
>field, designed to promote a certain hierarchical structure, motivated
>by perceived technological limitations.
>
>Other approaches are possible, but the consequences are thought to be
>bad -- for example, you could charge a single flat fee per address per
>year.  It is thought that the result of such a policy is that the
>backbone routers would melt, or something similar, and the internet
>would become unusable..



>OTOH, it would create much stronger pressures
>for improved router technology and perhaps accelerate a migration to
>IPv6.
>

Pressures for improved router technology certainly.  But it is questionable
if real alternatives, fully supported, are available in the short term to
help.

Longer term, of course--there are any number of approaches being explored.
But they are in the research and advanced development.

IPv6 is not inherently going to help the routing table growth problem.  It
helps many other problems, especially renumbering.  Provider change would
probably be much easier with fully evolved v6.

Howard