[arin-discuss] fee waivers

[Owen DeLong]

On Jun 25, 2010, at 11:48 AM, Ted Mittelstaedt wrote:

> 
> 
> On 6/24/2010 8:13 PM, John Curran wrote:
>> On Jun 24, 2010, at 9:53 PM, Ted Mittelstaedt wrote:
>>> 
>>> And also, IPv6 changes most of this since most ISP's are going to
>>> get the minimum IPv6 allocation and never come back for more numbers,
>>> thus I am contended for now with the fact that despite what ARIN
>>> does with regards to fees, in 20 years or so most ISPs will be
>>> paying the same amount.
>> 
>> How do you believe the fees should be structured 20 years
>> from now with respect to IPv6?  If one presumes that 98%
>> of the ISPs will be obtaining a single /32 allocation,
>> should the fees be the same for all of these organizations
>> or would you advocate a different model?
>> 
> 
> I would advocate the larger ISPs (larger meaning more users)
> pay more for the same exact
> reason the US Government's income tax system is progressive.

Interesting... I would advocate overhauling the US Government
Income Tax system and replacing it with a flat tax along the lines
of Malcolm Forbes "Fair Tax" proposal for much the same reason
I think the current system of fees at ARIN is reasonable and just.

> The larger operators gain far more financial benefit from the
> existence of ARIN to keep things orderly than the smaller operators
> do, just like people making a lot of money in the US have far
> more interest in a government that preserves order and puts
> thieves in jail and repairs the roads and all of that, then
> poor people who have nothing worth stealing and no car. :-)
> 
I actually don't buy this argument. Absent such a government, the
very wealthy have little trouble hiring their own private armed security
willing to do far more than the police are able to do to protect their
interest. In fact, statistically, people of lesser financial means are
far more likely to be victims of such crimes than people of significant
means and the government is rarely able to effectively prosecute
the kind of crimes that do usually befall people of means (mostly
committed by other people of means, amusingly enough).

> Under IPv6, ISP's are now going to be afforded a privacy about
> their internal operations that they never had.  Under IPv6 only
> I cannot calculate what any of my competitors sizes are because
> few to none of them will ever request multiple IPv6 allocations
> and thus I won't know if they are at 2% or 80% of their IPv6
> allocation.
> 
Good actors are still required to publish Whois records for their
delegations.

> Granted, I can query whois but because the whois database is
> such a mess right now, that data isn't trustworthy.  And even
> when it finally becomes trustworthy when ARIN finishes
> cleaning out the garbage POC entries, since ISP's today only

The good news is there isn't much IPv6 cruft in there yet, so, the
IPv6 cleanup should be relatively easy. The majority of the mess
is historic and IPv4 related.

> really feel the pressure to insert whois data in order to
> be able to justify for more IPv4 addresses, under IPv6 if
> they aren't ever planning on obtaining another IPv6 allocation
> it's easy to see that there will be little incentive for them
> to add records into whois.  So that data will be highly
> underreported except by the very largest ISPs who will be
> getting more than the minimum IPv6 allocations and will
> thus be forced to report accurately to meet the 80% rule.
> 
Perhaps we need policy allowing allocations to be reduced or
revoked if there are not enough SWIP entries to justify the
space, or, in cases where there is documented verifiable
underreporting?

> My guess is when most ISP's start to understand this, they will
> be extremely uninterested in changing it.  This is competitive
> data, ya know.
> 
Fortunately, ISPs are not the majority of orgs receiving resources
from ARIN these days. I do not know if they constitute a majority
of the membership any more, or not. Certainly it would be relatively
easy to overcome that fact if they do.

> ARIN is likely to see a big fee squeeze in the next decade. ISPs
> will not want to report utilization data to ARIN or to each other
> and will want to have everyone pay the same flat fee - but I
> suspect that in order to fund ARIN this flat fee will be even
> more regressive than the ARIN fees are now.  To maintain it ARIN
> will have to raise fees and that will push even more of the
> small ISPs out of the game.  The small ISP's will fight this and
> ARIN will be pressured to keep fees low - and since (as you I
> believe have pointed out before) the small ISP's have the voting
> power in numbers within ARIN, they will get their way.  Besides,
> we aren't dumb and we all know that Internet innovation comes
> from the smaller operators and ARIN would be doing a huge disservice
> to the Internet to do anything to push more of them out of business.
> 
If it is, as you say, then, the small ISPs could, theoretically make the
jumps in fees as you go up in size even larger.

However, I don't think most small ISPs feel the need to do this.

> I'd have to ask you this.  How many ISP's who are CURRENTLY running
> on an IPv4 Medium/Large/X-Large allocation will be able to switch
> to a IPv6 /32 Small allocation once IPv6 is in force - and once
> they start dropping their IPv4 allocations they will be moving

I would suspect very few given that I know of at least one ISP that just
transitioned from the IPv4 small to IPv4 medium to IPv4 large and has
nearly exhausted most of their IPv6 /32 and is looking at applying for
additional IPv6 space today.

> DOWN the fee scale.  How many customers can an ISP serve off of
> a /18 of IPv4 - and can they serve that same number off a /32 IPv6?
> I would think that they can, wouldn't you?
> 
Assuming business customers where /29 is the smallest practical
assignment (even if you just do it as 2 /30s for a point-to-point and
a CPE router+PC) you have 11 bits = 2048 theoretical customers
without allowing for infrastructure and other overhead.

A /32 gives you 65,536 /48 sized customers or 16.7Million /56
customers. A /32 really is equivalent to at least a /16, and, much
more like a /8 in the IPv4 world, but, IPv6 customers will be
inherently larger consumers of addresses, so, there's not really
a valid comparison there.

If I were to guess where things will settle out:

All x-small and small IPv4 ISPs will probably survive fine as IPv6 small.

Some medium IPv4 will probably be OK in IPv6 small, some will probably
need to be medium, a few may become large or even x-large.

Most large IPv4 ISPs will likely become IPv6 large or possibly x-large
in the near future, possibly even entering the xx-large category.

Most x-large IPv4 ISPs will be x-Large IPv6 ISPs fairly quickly, possibly even
xx-large.

I know the ISP I mentioned above is likely to become an X-Large IPv6
ISP on their next application to ARIN.

Owen