[ppml] Policy Proposal 2005-8: Proposal to amend ARIN IPv6 assignment and utilisation requirement - Last Call

[Owen DeLong]

>>
>> --On April 15, 2006 12:28:40 AM +0200 JORDI PALET MARTINEZ
>> <jordi.palet at consulintel.es> wrote:
>>
>>> Hi Owen,
>>>
>>> Yes, I know, it may seem so, but is not the case.
>>>
>>> Actually for the ISP is cheaper to keep a flat infrastructure, having
>>> everything with /48.
>>>
>> I think that's variable from an infrastructure perspective, however, from
>> an RIR fees perspective, if each /32 costs $x, then, each /48 costs
>> either $x/65536 or $x/n where n is the number of customers you can
>> put into said /48.  In the case where n>65536, the cost per /48 is
>> reduced.
>>
>>> I also don't think the RIR fees will make a difference, at least not in
>>> the short/medium term for that, but of course, this is something that is
>>> yet not clear.
>>>
>> Obviously at the moment when RIRs are waiving v6 fees, this is true.
>> However, given that a /32 costs $2,250/year and a /31 costs $4,500
>> per year, if I can multiply the customer count in the first /32
>> by as little as 2, it is useful.
>
> Yes, but following the same calculations, this means less than $0,035 per
> year per each /48 customer, so not a big deal, not an issue at all. I'm
> even happy to pay ten times that every month if needed, but I want to
> make sure that I don't need to explain to any ISP why I need a /48
> instead of a /56 and have the chance to get that rejected or need to look
> for an alternative ISP just because this.
>
I agree that the justification process should be limited to "my intended
topology requires it." up to a /48 per physical location (ignoring for
the moment the ambiguities in the definition of physical location).

However, as an example, in my house, I currently have 3 /24s which I am
utilizing efficiently according to ARIN policy.  I cannot envision at
this time a requirement for more than a /56 and would not ask my
ISP for more. (Frankly, a /60 would probably be plenty in my case).

I know there's stuff on the horizon that is planning to use my
refrigerator as gateway to the toaster network, coffee maker network,
microwave network, and oven network, not to mention networks for
the sink, the separate subnets for various controls in each bathroom,
and, of course the lightswitch network.  However, I just don't anticipate
those needing to be on anything other than SLA, so, I still don't
think I need more than a /56 from my ISP.

In my neighborhood, I am one of the more high-tech oriented houses
and I guarantee you the largest IP consumer in about a 1 mile
radius (which includes a couple of schools).

>> With the possible exception of free address pool fragmentation, I would
>> agree with you to a certain extent.  The only other problem I see is that
>> it won't be perceived as an issue until several ISPs have grabbed onto
>> multpile /32s with lots of empty reserved space and some other ISP(s)
>> need an unavailable /32.  I agree this is a long way off, but, in terms
>> of IPv6 potential useful life, a long way off might not be far enough.
>
> Statistically seems to me highly improbable, because even in a situation
> of high level of addressing space utilization (which will never happen
> because we will have done a policy change before going into that
> situation), new ISPs will only happen if other close the business or
> similar situations. Anyway, is a complex situation to predict, and for
> sure we will not have the right crystal ball to look at and make a good
> guess ;-)
>
That simply isn't true.  There are new ISPs in the US on a nearly daily
basis.  Many of them close their doors or are acquired in 1-2 years,
but, they spring up constantly.  Unfortunately, in the case of acquisition,
either the acquiring organization will either acquire the address space
as well (minimal disruption to customers and most likely scenario)
or require some period of time to renumber customers out of the space
(usually measured in multiple years in my experience).  In either case,
the churn in the industry alone is likely to result in address space
fragmentation.

>>
>>> I think we should remember that if we agree to change the HD-ratio, the
>>> figures that Tony has calculated, and I think are realistic, give IPv6 a
>>> life of 480 years. Anyone still believe that IP (IPv4 or IPv6, doesn't
>>> matter for this case), will be still available in 200 years ? So ?
>>>
>> I've seen numbers (I think from Geoff Huston) that said it was more
>> like 60 years.
>
> I think Geoff figures were assuming that no change is done in the
> HD-ratio, right ?
>
That wasn't my understanding, but, I could be wrong.

Owen


-- 
If this message was not signed with gpg key 0FE2AA3D, it's probably
a forgery.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 186 bytes
Desc: not available
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20060415/aaf22d64/attachment.sig>