[arin-ppml] Draft Policy 2011-5: Shared Transition Space for IPv4 Address Extension - IAB comment

[Owen DeLong]

On Jun 28, 2011, at 10:35 PM, Joel Jaeggli wrote:

> 
> On Jun 28, 2011, at 5:38 PM, Jimmy Hess wrote:
> 
>> On Tue, Jun 28, 2011 at 9:50 AM, Joel Jaeggli <joelja at bogus.com> wrote:
>>> On Jun 28, 2011, at 5:50 AM, Jimmy Hess wrote:
>>>> [snip]
>>>> What assumptions would those be?
>>> That a port mapped to a the outside of a cpe which does not have an rfc 1918 address will in fact be reachable (example by upnp or nat pmp)
>> 
>> We don't need to address all the issues with NAT and NAT444 over again
>> in this discussion;  there are multiple RFCs and drafts discussing NAT
>> and its issues,  let's concentrate on any _additional_ assumptions
>> that are violated by ARIN allocating a shared address space
>> that are not violated by the mere use of NAT444.
>> 
>> This is basically an argument against the use of NAT444 ever,  but if
>> ISPs want to utilize NAT444, this assumption is violated whether a
>> special shared range is allocated or not;
> 
> no it isn't. there is a tangible difference between a scenario where something will not be configured and therefore will never fail and a, something which fires packets into the ether becuase it doesn't know any better. LSN is here, allocating a new prefix or squatting on public scope prefixes will break some fraction of old cpe (it does today) in worse way then allocating them out of private scope prefixes (for which they already have logic) until they age out of the network.


Due to the support issues, no provider in their right mind is going to use 1918 space for the middle-layer in a NAT444 scenario.

As such, the question remains the same...

Do we allocate a single /10 so that everyone can use the same address space, or, do we force each provider to build their own collection of GUA addresses allocated to this same purpose?

Owen