[address-policy-wg] 2013-03 New Draft Document and Impact Analysis Published (No Need - Post-Depletion Reality Adjustment and Clean up)

Filiz,

> According to the current policy your main reason to request an IP 
> address must be that you need it to use on a network.

This is only the true for *assignments*, not *allocations*. This is a
subtle, but important, difference.

> As soon as you remove this, that the IP address is actually needed
> on a "network" (network can be LIR's or End Users, does not matter),
> you make them a commodity. As soon as you do that within your policy,
> you stamp that they actually are money making assets to be turned
> into some value now or later and this practice is all fine according
> to the policy too.

We have *today*, under our current policy:

- An allocation transfer policy that has no restrictions on monetary
compensation being exchanged between the parties to a transaction;
- A list of «Recognised IPv4 Transfer Brokers» published by the NCC;
- A NCC-operated "lightweight self-service IPv4 broker" or "bulletin
board" aimed at helping LIRs interested in trading find each other.

2013-03 takes no position whether or not all of this is a "good thing",
nor do I. But let's face the facts, it's *already there*.

> This means an LIR can request these resources to do whatever they
> want to do with them including requesting them not to be used on any
> network of their own or an End Users' at all but for some other
> reason. One of these reasons will be deliberately passing them on to
> a paying 3rd party who may or may not use them on any other network
> either.

This is *already possible* today; in fact, what you're describing above
is pretty much the core function of an LIR.

Let me try to explain by making an hypothetical example:

Background: John Q. Public wants to make money on IPv4. He has a laptop
and and e-mail account, but no networking equipment, nor an internet
connection of his own; in fact, he's leeching off his local coffee shop
gratis WiFi in order to get online. John then does the following:

1) Incorporates "JQP IPv4 Leasing Company" in his local jurisdiction.
2) Joins the NCC and pays the membership fee, thus becoming an LIR.
3) Requests and receives an IPv6 allocation, ignores it.
4) Requests and receives his "last /8" /22.
5) Sets up an auction or something like it, and finds the end user(s)
willing to pay the most for [parts of] the /22.
6) Gets a completed ripe-583 form from the auction winner(s), forwards
it to the RIPE NCC (if he does not yet have an AW), and saves a copy in
an archive folder.
7) Assigns the address space over to the new end user(s) (i.e.,
registering ASSIGNED PA inetnum objects in the RIPE database).

So, again, this would be *already possible* and completely legitimate
under our current policy. If John manages to get higher income from his
customers then what he's paying the NCC in membership fees, then voila!
John is making money off IPv4, without having ever forwarded a single
IPv4 packet on behalf of his customers.

Again, 2013-03 takes no position whether or not this is a "good thing",
but it is the status quo today, so when discussing the proposal, I think
we should try to keep on-topic, i.e. discussing what the proposal
actually *changes*, which in the above case is the removal of point #6
and nothing else:

John's customer won't have to fill out the ripe-583 form anymore. This
is a good thing at least in one way, because it removes a bureaucratic
overhead that would otherwise impose a work load on John, John's
customer, and the RIPE NCC. I think you referred to it as "admin
hogging" in an earlier message and I couldn't agree more.

Your concern is that by removing point #6, it becomes possible that the
winner(s) of John's auction does not "need" the space, yet are able to
receive it. I concede that this is theoretically possible. But I remain
unconvinced that this is an *realistic* problem - I simply don't see why
someone would bid on address space that they don't "need", especially
when they have to outbid a bunch of others who *do* have potentially
quite desperate "need".

And even if such actors do exist and do outbid all the folks that do
"need", keep in mind that *today* 1) the entire system is trust-based
and that 2) John has absolutely no incentive to undertake a deep
investigation into whether or not his customer's ripe-583 form is
truthfully filled out or not - if it looks credible enough, his job is
done. He could also make a habit of forwarding the forms to the NCC to
get "official" stamps of approval.

> I do not think this is good practice or responsible address 
> management.

Maybe, maybe not - but it is what we have today. If we want to ban paid
IPv4 allocation transfers and assignments because it's these do not
constitute "good practice or responsible address management", then we as
a community can do that, but please let's keep that discussion in a
separate proposal.

> And so in fact I am curious what is the rush of this proposal now 
> that RIPE NCC is allocating from the last /8 and that soon it will
> be exhausted anyways and what is wrong with keeping what we have as 
> policy as long as this last bit of /8 lasts.

Your definition of "soon" differs wildly from mine, or you haven't done
the math. Allow me:

- Number of /22s in 185.0.0.0/8 (sans the /16 for IXPs): 16320
- Number of LIRs that held allocations outside 185/8 at the time of
depletion (these are the only ones who can get their last /22 as an
additional allocation): 7912
- Number of reclaimed space held by the NCC as of today, in /22s: 903
- Number of reclaimed space held by the IANA as of today, divided by 5
RIRs, in /22s: 3731

In other words, there is a total of 13042 /22s to be given out to new
LIRs under the last /8 policy (and this number keeps increasing as the
IANA and the NCC continues to recovers space). As of today, there are
959 LIRs who hold *only* a /22 from 185/8. 314 days have passed since
IPv4 depletion.

So how long will the last /8 last us based on the current usage rates
and available space?

13042 / (959 / 314 * 365) = 11.7 years

I truly hope that we've gotten IPv6 properly out the door by then...

To answer the second question, "what's wrong with keeping what we have":
About half of the current policy text is describing out of date
concepts, is self-contradictory or otherwise wrong, and that it imposes
a bureaucracy on the community that demands that hundreds of thousands
forms be filled out every year, all in the name of delaying something
which has already happened.

I mean, you even said it yourself in an earlier message: «Yes, agreed,
this is bureaucratic, admin resource hogging etc and this needs a
change» - and I could not agree more, this is *exactly* why I have
proposed 2013-03.

> Finally Gert seems to get my point with his example of someone going 
> through the effort of opening 5000 new LIRs. I am not saying this 
> will happen but even someone going to through the effort of opening 
> maybe 2 or three just to get more address space for the wrong
> reasons should be telling us something that our policy is missing
> some core notion, that notion being IP addresses are mainly there to
> be used on networks as soon as they leave an RIR pool.

I believe Gert's point was that you can open 5000 LIRs in this manner
*today*, under our *current* policy. 2013-03 brings no change.
Therefore, this is not a valid argument against 2013-03.

If you want to close this loophole, then that's food for another policy
proposal. Opposing 2013-03 won't do anything about it, one way or the
other (nor would supporting 2013-03).

(That said, in the current data there's no evidence of someone
attempting to use this loophole in the described manner. Personally I
believe that the NCC's membership fee is a sufficient deterrence against
it.)

Tore