[arin-ppml] 2016-3 Revisited - anti-abuse clause
Jason Schiller
jschiller at google.com
Thu Feb 9 12:45:14 EST 2017
Maybe fairness was not the word I was looking for...
I don't mind large organizations using a more difficult process or
requiring greater proof, as was the case with the "less simplified
provisions of the existing policy" wrt the ISP slow start policy of the
ARIN pool. But slow start as it is currently written does not work well in
the transfer market. So instead the specified transfer is granted based on
some hand wavy, future looking prediction with unpredictable results.
Everyone should be able to use a process that is predictable, and not
dependent on hand wavy future looking projections.
For ARIN Pool space, as an ISP, I could "slow start" if I have no history
of utilization, or if my history of utilization is less than my current
growth rate.
In the case of the former I start with a small block (initially /22, later
between a /24 and /22).
In the case of the later, I start with a block size based on my previous
year's run rate (initially a 2 year supply, then 3 month supply, and now
again a 2 year supply).
I use up what I got. If I can use it in less time than the time window of
space I am allowed to get I can double, if it takes the time window, I get
that size again, and if it takes longer than the time window it goes down.
(plus rounding up initially or down later.)
This approach is:
1. predictable
2. is based on A. actual demonstrable utilization, and B.the belief that
growth will continue
3. does not require product folks to make some sort of hand wavy future
projection
4. does not require ARIN to engage in back and forth in an attempt to
assess the veracity of the projection
5. does not depend on a future looking belief which in the end cannot be
tested, proven or disproven.
6. does not depend on a claim that if not met can simply be excused with
"well, we missed" with no harm no foul.
7. does not reward more overly optimistic organizations with a greater over
supply.
My desire is to permit everyone (end users and ISPs) to use a predictable,
slow start mechanism in the transfer market.
My suggested transfer text differs essentially in three ways:
1. Demonstration of past utilization is slightly easier.
For ARIN IP space you have to show 80% utilization on average, and 50%
on each block.
For the suggested text its 80% on average and growth equal to 50% of
the space transferred.
- the 50% clause is to prevent someone with high utilization from
getting IP space, not using it
and their now lower utilization still being high enough to get more IP
space.
- the problem with this approach is someone may actually be growing,
needing, and using the new
space but have a small historically under utilized block, say a
deployment that got delayed.
- the growth equal to 50% of transferred space clause it to meet the
intent of the policy while giving
flexibility to renumber
2. Historical growth may not be a good measure because growth may be rate
limited waiting on a transfer.
- Basically the first block is "easy".
- We have a policy to let those with no space to get a /24
- For those with efficient utilization they can easily double (but not
bigger than a /16).
If the org never grows it could have up to a 50% surplus after using
this policy.
- Any org growing at less than the doubling rate will have trouble with
the 80% on average re-certification,
and will not be able to get another block.
They could have up to a 50% surplus if growth stops shortly after using
this policy
- Any org growing at more than the doubling rate needs and is using the
space, and should be able to get more.
- An org's who is doubling in size, will have trouble with the 80% on
average re-certification, when the growth
rate drops below doubling.
They may have up to a 50% surplus if growth stops right after they use
this policy for the last time.
- An org's who growth rate is greater than a /16 over the time window
(currently suggested as 6 months)
may have up to a /16 surplus if they if growth stops right after they
use this policy for the last time.
3. An organization will also have to a pay fee for the IP space
- this is likely a significant deterrent for small organizations
On Wed, Feb 8, 2017 at 3:35 PM, Owen DeLong <owen at delong.com> wrote:
> Respectfully I reject your premise on the fairness.
>
> Neither A, nor C prevent large organizations from getting more, they
> merely require that they use other less simplified provisions of the
> existing policy.
>
> I think what I support is sort of a hybrid between A and C in that I
> believe you should be able to use the policy to transfer as often as you
> want, so long as your transfers within any 6 month period under this policy
> don’t exceed a /16. You’d still be able to transfer a /16 under this policy
> and then use other existing policies if you needed more.
>
> Owen
>
> On Feb 7, 2017, at 12:04 , Jason Schiller <jschiller at google.com> wrote:
>
> I support B.
>
> It puts added work on those who need more than a /16, or have a growth
> rate more than doubling every half yeah, but does not prevent organizations
> who need IP addresses from getting them.
>
> I oppose A and C as they are unfair,
>
> A.
> - unfairly penalizes large organizations that need more than a /16
> - unfairly penalizes organizations growing faster than double their
> current holding
> (especially new organizations that started with a /24 and have a
> growth rate greater than 512 customer per year)
>
> C.
> - unfairly penalizes large organizations that need more than a /16
> - unfairly penalizes organizations growing faster than double their
> current holding
> - unfairly does not penalizes organizations growing faster than double
> their current holding so long as they need less than a /16
>
>
> A > B or B > A?
>
> I can't decide if A is less unfair because there is no carve out for
> organizations that need less than a /16. On one hand those needing less
> than a /16 are not treated unfairly as a special class, but as a result the
> number of organizations who need IP addresses that are rate limited is
> greater.
>
> Or if C is less unfair because it is unfair to have a carve out for the
> organization that need less than a /16 for exactly the opposite reasons.
>
> __Jason
>
> On Tue, Feb 7, 2017 at 2:53 PM, Jason Schiller <jschiller at google.com>
> wrote:
>
>> We have a few options on the table and only a few voices in the
>> discussion...
>>
>> I'd like to quickly outline the options, and see if we can get more
>> people to weigh in and either note they object to one or more options, are
>> ambivalent to one or more options, or support one or more options (with
>> some preference).
>>
>>
>> 1. demonstrate 80% utilization on average for all your IP space
>> 2. get pre-authorization for 1 or more transfers up to double your
>> current holdings over then two years
>> 2.1. this is limited to a /16
>>
>> A. you can use this policy once every 6 months
>>
>> B. If you need to use this policy more than once every 6 months you need
>> to also demonstrate growth equalling half what you have transferred since
>> you last used this policy.
>>
>> C. you can use this policy to transfer a total of up to a /16
>>
>> Where do you stand on A, B or C?
>>
>> __Jason
>>
>>
>> On Fri, Feb 3, 2017 at 7:01 PM, Scott Leibrand <scottleibrand at gmail.com>
>> wrote:
>>
>>> That would be a significant improvement on the current ("An organization
>>> may only qualify under 8.5.7 once every 6 months.") text. I would be
>>> equally fine with this text ("No more than a total of a /16 equivalent
>>> may be transferred under these provisions within any 6 month period." or
>>> similar) or with Jason's ("An organization may only qualify under 8.5.7
>>> once every 6 months, unless they can also demonstrate growth of IPv4
>>> utilization of at least half of the amount of specified transfers since the
>>> previous transfer pre-authorization or approval.")
>>>
>>> Thanks,
>>> Scott
>>>
>>> On Fri, Feb 3, 2017 at 2:22 PM, Owen DeLong <owen at delong.com> wrote:
>>>
>>>> Simple to resolve for the 6-month horizon —
>>>>
>>>> … Such that no more than a total of a /16 equivalent may be transferred
>>>> under these provisions within any 6 month period. …
>>>>
>>>> Owen
>>>>
>>>> > On Feb 3, 2017, at 07:19 , David R Huberman <daveid at panix.com> wrote:
>>>> >
>>>> >
>>>> > I thought of a possible problem with the anti-abuse language -- all
>>>> versions of it. Let me talk it out.
>>>> >
>>>> > An organization has a /19.
>>>> > It has growing products, and wants another /19 for its 1 or 2 year
>>>> need.
>>>> > It wants to avail itself of the new language.
>>>> > It is able to buy a /20 from Buyer A, and a /20 from Buyer B.
>>>> >
>>>> > It closes the deal with Buyer A first, and transfers at ARIN using
>>>> the proposed language.
>>>> >
>>>> > How does it use any version we've discussed (Jason's various
>>>> proposals, the current text, etc) to transfer the space it buys from Buyer
>>>> B?
>>>> >
>>>> >
>>>> > (In all discussion, yes, you can always use the other sections of
>>>> 8.5, but let's stick to the spirit of this policy language, which is meant
>>>> to help smaller and mid-size networks double their holdings without needs
>>>> testing.)
>>>> > _______________________________________________
>>>> > PPML
>>>> > You are receiving this message because you are subscribed to
>>>> > the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
>>>> > Unsubscribe or manage your mailing list subscription at:
>>>> > http://lists.arin.net/mailman/listinfo/arin-ppml
>>>> > Please contact info at arin.net if you experience any issues.
>>>>
>>>> _______________________________________________
>>>> PPML
>>>> You are receiving this message because you are subscribed to
>>>> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
>>>> Unsubscribe or manage your mailing list subscription at:
>>>> http://lists.arin.net/mailman/listinfo/arin-ppml
>>>> Please contact info at arin.net if you experience any issues.
>>>>
>>>
>>>
>>> _______________________________________________
>>> PPML
>>> You are receiving this message because you are subscribed to
>>> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
>>> Unsubscribe or manage your mailing list subscription at:
>>> http://lists.arin.net/mailman/listinfo/arin-ppml
>>> Please contact info at arin.net if you experience any issues.
>>>
>>
>>
>>
>> --
>> _______________________________________________________
>> Jason Schiller|NetOps|jschiller at google.com|571-266-0006
>> <(571)%20266-0006>
>>
>>
>
>
> --
> _______________________________________________________
> Jason Schiller|NetOps|jschiller at google.com|571-266-0006 <(571)%20266-0006>
>
>
>
--
_______________________________________________________
Jason Schiller|NetOps|jschiller at google.com|571-266-0006
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20170209/0ec0fbf8/attachment.htm>
More information about the ARIN-PPML
mailing list