I didn't have any popcorn but a few nachos were helpful to read the full e-mail. Very good and detailed explanations. +100 from me to Elvis which can also be read as -100 for the policy. For those of you who pretend working, it's friday so you can't trick anyone ;). You'd better read Elvis's mail entirely. Have a nice weekend, Ciprian On Fri, Oct 21, 2016 at 4:36 PM, Elvis Daniel Velea <elvis@velea.eu> wrote:
Hi,
On 10/19/16 11:05 AM, Marco Schmidt wrote:
The goal of this proposal is to ban transfers of allocations made under the final /8 policy. Also the proposal specifies what resources must be added to the RIPE NCC IPv4 available pool.
I do not agree with this policy proposal and believe it will not fix anything, instead - it will harm the registry.
Some of the differences from version 2.0 include:
- Clarification that changes to holdership of address space as a result of company mergers or acquisitions are not affected by proposed transfer restriction
this fixes only a tiny bit of the problem.
- Legacy space handed over to the RIPE NCC will be added to the IPv4 available pool
this has nothing to do with the policy proposal. I feel it's just some candy offered to sweeten the proposal itself.
This was the short version of my response. Those reading this e-mail during working hours, you can go back to work ;) those that still have some popcorn left, feel free to read further.
Below are the 6 most important reasons why I believe this policy proposal should not become policy:
1. This policy proposal will create two types of members.
a. the members that have received resources before 2012 + the members that can afford to 'buy' IP addresses allocated until recently (-2y from the date this policy proposal would be implemented) b. the members that have only received resources after September 2012 and can not afford to buy IP IP addresses at the market prices (but they can buy an unlimited number of these from the RIPE NCC at ~€4,5 (€3,4/1st year + €1,4/2nd year - redistribution of profit)
The first type of member would be allowed to participate in an IP transfer market that was (until the implementation of this policy proposal, if ever) accessible to everyone and anyone with resources received from the RIPE NCC or an other member. The second type of member will not be allowed to participate in this IP transfer market unless they buy first from an other member.
Some members have already been able to transfer their /22 received from the last /8. With the implementation of this policy proposal (if ever) I feel that we as a community will discriminate between those that have received their last /22 (and want/need, for various reasons to transfer it) 2 or more years before implementation and those that have received less than 2 years before the implementation (or any time after).
I know some of you do not like analogies. But I would compare this with 2 people buying their cars. The first can buy the car and sell it 2 years later, only because the purchase happened even just a few days before the car industry decides that cars can no longer be sold further. So, those that have bought their car and used it for 1.5 years will have to return it (for free) to the car manufacturer while others could have sold it because they were quick in the purchase process and bought it earlier.
So, I think that once this policy proposal would be implemented (if ever) it would discriminate the second type of member as they will not be allowed to participate in a well established IP transfer market with the IP addresses received from the RIPE NCC (and only with the IP addresses they need to buy from the market first).
2. This policy proposal will create yet an other color of IP addresses.
I believe and hope that in the near future we will start talking about the removal of colors and not about addition of more colors. Numbers are just numbers. There is no difference between a number received in 1990, one received after 1992 or one that has status PI or PA. Now, we want to add a color for numbers received after 2012...
My router does not know any difference between these numbers and nobody really cares. Do you think that PI holders care that they are not supposed to sub-assign that space to other customers ? Do you think the RIPE NCC has any say or can really verify who (and most importantly - how someone) is using any of these numbers? The community decided to remove most of the barriers when 2013-03, 2014-02, 2014-05 were approved and implemented. These policy proposals removed all the 'old' requirements and cleaned up the IPv4 policy so much that anyone can now do whatever they want with their IPv4 space (not that they could have not done it in the past, but they would have to lie to the RIPE NCC if they would have ever needed an additional allocation). This policy proposal tends to take us back to previous way of thinking, an old one - I think.
So, just adding more colors and barriers will only complicate things. And I've always liked policies and procedures that are clear and simple. As simple as possible.
3. This policy proposal will drive some IPv4 transfers underground.
We already know and have seen it in previous presentations that where the RIR system tries to impose some limitations, the real world invents something to circumvent those limitations. Currently, with very large allocations, the invention is called a 'futures contract'.
While the RIPE NCC policies will say that ALLOCATED FINAL can not be transferred between LIRs (if this policy proposal will ever make it into policy), most LIRs will find ways to circumvent the policy (as they have always done it). They will create contracts that will stay underground and will be covered by NDAs, they will use loopholes in the M&A process, they will do whatever they need to in order to (sell/buy) the resources they needed.
We've seen that some of the people that participated in the previous discussion of this policy proposal said they will never do such a thing, they will never buy a /22 that says ALLOCATED FINAL. Would you change your mind if your business depended on it? If there was no other way to receive an IP block that you desperately need ? If you have a very important customer that would leave if you can not assign a few more IP addresses? I bet anyone that has a commercial business will think twice and examine again and again where is that border between profit and loss and whether respecting the RIPE Policies to the very last letter really matter that much that they would rather have a loss.
4. This policy proposal will drive the free IPv4 pool to run-out even faster.
What will the LIRs do if they won't be able to find affordable IPv4 in the market any longer? If these /22s that were allocated from the last /8 are no longer available in the transfer market? Well, they will either buy more expensive (pre-2012) IP addresses or go to the RIPE NCC and receive a /22 for less than €5/IP. The board announced publicly that anyone can open multiple LIRs if they need more IPs.
Approving this policy proposal will only remove some 'assets' from the free market and the supply will shrink, therefore driving prices up for whatever is pre-2012 and pushing more members to get the IPs they need from the RIPE NCC by means of setting up multiple LIRs.
5. This policy proposal will affect the registry (in a bad way)
As said in #3, some transfers will be driven underground. We do not see many of those in the RIPE Region at this moment because there is no limitation on who can transfer and when (except for the two years anti-flip bit). If this proposal ever becomes policy, the transfers of the /22s will still happen. If the two parties can not convince the RIPE NCC of their 'M&A' and can not get the IPs changed in the RIPE Database from a name to an other, they will keep on using them with the wrong name.
This will affect the quality of the registry and we will have a registry with two colors. The bleached IPs will be the ones pre-2012 and the rest will have various colours of grey.
6. Artificial explosion of number members
The implementation of the members' vote (on the board decision of blocking the registration of multiple LIRs on the same company name) requires those that need a second /22 to keep it in a second LIR for at least 2 years before this second /22 can be transferred to their first LIR. The second LIR can then, after the transfer, be closed. So the price per IP, has been set to a minimum 2 years membership + sign-up fee - redistribution.
It may actually require the company opening multiple LIRs to keep all of these open if they want to hold on to their IP addresses. It is not very clear what will happen when these companies will want to consolidate all the LIRs they have into one. If they will be forced to return the /22 they got on their second LIR, then they will be forced to keep this second LIR open.
If that's the case, the number of LIRs will explode and will not shrink every time a second LIR ages 2 years and can transfer the /22 to the first. I am unsure whether Remco, as a private citizen but more, as a Board member has given this enough thought. I believe that this policy proposal will make the RIPE NCC more and more unstable with difficulties in making financial plans on the long run.
I worry that these companies that will be forced to keep multiple LIRs open just to hold on to their /22s may take the Board or the NCC to court because they are not allowed (after 2 years) to consolidate their membership into one registry. I am unsure what a court would say when they are told that Remco has multiple hats and his proposal is not made with a Board hat. To most of the world (and I've talked to quite a few people about it), this policy proposal seems to come from the RIPE NCC Board. To many people that I talk to, this policy proposal is seen as a double standard from the NCC Board. On one hand, the Board decided to advise the members to vote to allow companies to open multiple LIRs. On the other hand, one of the Board Members (even if as a private citizen) makes a policy proposal that forces those to pay a la longue for each /22 they received.
There may be other reasons why I do not like this proposal. The ones mentioned above are the most important ones.
Regards, Elvis