[address-policy-wg] can deadbeat LIRs reverse IPv4 exhaustion?

On Thu, 7 Feb 2019, Jim Reid wrote:

>
>
>> On 7 Feb 2019, at 07:59, Carlos Friaças via address-policy-wg <address-policy-wg at ripe.net> wrote:
>>
>> Even when the pools reach ZERO, if 1000 LIRs stop paying fees (and that's only one example/route), the "runout" will be temporarily reverted, and handing out IPv4 addresses will be again, in theory, possible.
>
> How is that possible? Once the pools reach zero, there are no more addresses to hand out.

At that point in the timeline, YES. zero means zero.


> An RIR can't conjure up IPv4 address space out of thin air. If it was 
> able to do that, we could just continue forever with business as usual 
> and allocate v4 until the heat death of the universe.

Yes, that's correct. But a set of foreseeable events might pour down 
-some- IPv4 space, growing the stock from zero.

The NCC registration services tell us they are getting addresses back 
*every* year (yes, that was a surprise for me too).
Even if that doesn't happen during a full year, it doesn't mean it won't 
happen in subsequent years. If i didn't get it wrong, that depends on a 
variety of factors.

Of course the "yearly recovered numbering assets" are not enough to cope 
with all the demand -- that's when the waiting list might be useful...


> Besides, there?s no mechanism or policy for the NCC to recover 
> addresses from LIRs that don?t pay their bills.

I think you are wrong.

Apart from the financial side, if a LIR doesn't comply with policies 
(falsified data, and so on...) there is a service termination process and 
resources go back into the pool after some time -- please someone at the 
NCC, tell me if i got it wrong.


> If such mechanisms or policies existed, they?d be unworkable. There?s no 
> way of knowing for sure that those addresses weren?t being used.

Bad luck. Rules breaking means revokation...
The higher risk (as i see it) goes to new recipients of the space, after 
some quarentine.


> So if they were reclaimed, the addresses couldn?t be allocated to 
> someone else.

I think the NCC and current policy might disagree -- please tell me if i'm 
wrong.


Best Regards,
Carlos