[ppml] Policy Proposal: Decreasing Exponential Rationing of IPv4 IP Addresses

[Iljitsch van Beijnum]

On 20-aug-2007, at 16:29, Member Services wrote:

> Policy Proposal Name: Decreasing Exponential Rationing of IPv4 IP  
> Addresses

I'm opposed to this policy.

The main reason is because its premise is flawed: the policy supposes  
that making sure we don't run out of IPv4 address space the next 10  
years is better than the situation where we do run out. However,  
addresses that sit in the ARIN warehouse unused don't do the  
community any good, while more stringent rules are harmful, because  
they make new deployments harder, take longer, and increase risks.

Also, the policy doesn't describe a workable practice for the ARIN  
staff. What if 8 requests come in, 4 for 1 million addresses and 4  
for 250000, and the current rationing is for 4 million? Do 4 people  
get their million and the 4 others get nothing? Or 7 get everything  
and one gets nothing? Or they all get 80%? In that case, the people  
who need 250000 will obviously ask for 400000 so they get 285000 but  
the big ones only get 700000. Or do the small ones get everything  
they ask for and the big ones only 75%?

If you only get a percentage of the address space you need, this is  
nearly as harmful as not getting anything at all. And the benefits of  
gaming the system increase astronomically, so "address request fraud"  
will become much more common.

As I said, rationing IPv4 addresses is a bad idea. But if the  
community wants to do it, there are better ways.

Besides, something similar to rationing will happen by itself: around  
10% of all requests is for 90% of the yearly address space used,  
meaning that it's for very large address blocks. At some point, these  
blocks will no longer be available. But the other 90% of the requests  
that only use up 15 - 20 million addresses a year can still be  
fullfilled from smaller blocks that remain available and from address  
space that is reclaimed. (This is more than 10 million addresses per  
year.)

So small amounts of IPv4 address space will remain available in some  
form for years to come, the "running out" moment is the point where  
the largest requests, done but the world's largest ISPs, can no  
longer be fullfilled.