[arin-ppml] [Fwd: Draft Policy 2011-5: Shared Transition Space for IPv4 Address Extension]
Owen DeLong
owen at delong.com
Tue Feb 22 03:58:11 EST 2011
On Feb 21, 2011, at 6:24 PM, Joel Jaeggli wrote:
> On 2/21/11 5:37 PM, Owen DeLong wrote:
>> RFC-1918 is not an option because it overlaps space in use by the end
>> users for their local area networks.
>
> So you say, I say you have the choice between two support problems, one
> which you don't have presently but many isps the wold over contend with
> every day.
>
Care to explain what these are?
>> It may not invalidate their use, but, it
>> does make it impracticable from a service provider perspective.
>
> This is an opinion, not a fact. It is a fact that collisions between the
> external address assigned and the internal addresses utilized will
> render some percentage of cpe inoperable. By some accounts the amount of
> cpe that both have that problem, and will be numbered in such a fashion
> thereby forming the union that causes that problem is a fraction of the
> total population.
>
That union in a very large user population (such as faced by many
of the large residential ISPs in the ARIN region) is large enough to
be potentially quite costly to support.
>> Definitely
>> much better to look for an un-advertised range and squat there from a
>> pragmatic perspective.
>
> Not if an application, be it reliant on upnp, 6to4 or some other
> mechanism for leveraging the external address assumes the external
> address on the cpe is global in scope.
>
That application will break in this scenario anyway, regardless.
That assumption is going to end up being false in the near future
no matter what.
> The assumption that an unadvertised range will in fact remain so, has
> been repeatedly and routinely proven to be false and ISPs that have done
> that do so at their peril...
>
Only if they think some percentage of their customers might want to
reach that specific part of DoD when/if they start advertising it.
I would say at this point, for the likely lifetime of IPv4 and especially
NAT444, that's reasonably low risk compared to the support costs
inherent in using RFC-1918 in this situation.
However, you are correct that it isn't the best solution. What I expect
will happen is that they will, instead, each seek large allocations from
ARIN to support their NAT444 intermediary addresses and get them
under current policies, thus accelerating IPv4 exhaustion.
>> This is explained if you read the proposal in its entirety.
>
> I have, I also have read previous iterations of the proposal, both here,
> in the APNIC region and in internet draft form at the IETF.
>
> As a network operator I opposed the 127, not because because I believe
> that there isn't a problem but because I believe that the proposed cure
> is worse than the disease...
>
> If you must hand out private scope addresses do so. triage small
> percentage of cpe that can't reach your gateways and move on.
>
Triaging small percentage when you have 20 million subscribers
may be a rather large triage process.
> At a minimum the proposal should acknowledge that we're trading one kind
> of breakage (private scope v4 address) for another (presumed to be
> public scope v4 address but actually private) and that alternative 4 is
> in fact the thing that you don't want to do which is just use rfc 1918
> and deal with those consequnces which are well understood...
>
A provider that considers it practical to do so is certainly welcome to use
RFC-1918 and there is nothing in this proposal that precludes that.
However, for a certain percentage of providers, that is not practicable
in their environments due to rather enormous support costs that would
result. In those environments, the likely choices the provider will make
will be between getting this policy so they can all share the same
semi-private /10, or, each getting their own separate allocations which
may be as much as a /10 in some cases and will almost certainly add
up to way more than a /10 in toto.
Owen
>> Owen
>>
>> On Feb 21, 2011, at 4:58 PM, Joel Jaeggli wrote:
>>
>>> The following statement from 2011-5 is incorrect or at a minimum chooses
>>> to deliberately rule out one option.
>>>
>>>> Service providers are currently presented with three options for
>>>> obtaining sufficient IPv4 address space for NAT444/IPv4 extension
>>>> deployments: (1) Request allocations under the NRPM; (2) share address
>>>> space with other providers (this proposal); or (3) use address space
>>>> allocated to another entity (i.e. ‘squat’). Of the three options,
>>>> option 2 (this proposal) is preferable, as it will minimize the number
>>>> of addresses used for IPv4 extension deployments while preserving the
>>>> authority of IANA and RIRs.
>>>
>>> Which is use RFC 1918 space. The fact that there are conflicts with
>>> addresses used in gateways in no way invalidates the suitability of
>>> private scope ip addresses for use in a private scope. Creating new
>>> private scope ranges which gateways do not treat as such has it's own
>>> liabilities and at a minimum that needs to be acknowledged and balanced
>>> against threat of collisions.
>>>
>>> joel
>>> _______________________________________________
>>> 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.
>>
>>
More information about the ARIN-PPML
mailing list