I don't agree with the current policy. Reasoning : As Gert already stated: companies will find ways around this anyway by selling parts of their companies / stocks which would have the IP’s otherwise. I already heard of a situation where an Asian based company opened a UK ltd just to become a RIPE LIR member in order to be able to apply for IP in our region. Bottom line is. People with a specific goal are cunning and more creative than what we allow or not in a specific policy. Especially if (a lot of) money is involved. However by releasing this policy, (semi)authorizing a market that would prolong IPv4 lifetime and further delay v6 deployment, would give the wrong signal to the public in my personal opinion. Any kind of transfer of unused space, should go back to the RIR and not end up with the highest bidder. The sooner people realize that v4 is soo 2000, the better for us all. Regards, Erik Bais
Any kind of transfer of unused space, should go back to the RIR and not end up with the highest bidder.
I agree. If we really feel we have the need to transfer IPv4 address space between RIRs, the transfers should happen between RIRs, not LIRs. I'm not so sure that is a good idea either... Also, we already have a policy for transferring address space within the RIPE-region. Why is this proposal trying to create a new policy instead of amending what we already have in place? Having drastically different rules for different kinds of transfers is confusing, resulting in something similar to IPv4 PI versus IPv6 PI. ____________________________________ Tero Toikkanen Nebula Oy
Hi, On Fri, Apr 13, 2012 at 06:57:16AM +0000, Tero Toikkanen wrote:
Also, we already have a policy for transferring address space within the RIPE-region.
"within", yes.
Why is this proposal trying to create a new policy instead of amending what we already have in place?
Because our current policy doesn't say anything about transfers outside of the RIPE region (plus, our current policy also needs work, but that's a different thread). Out-of-region transfers are (obviously) different because the rules of the corresponding RIR need to be taken into account. Gert Doering -- APWG chair -- have you enabled IPv6 on something today...? SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
Hello, I don't have a very strong opinion about this, but i think i'm on the opposing side to this proposal. I may be wrong, but i guess this community has some rules about returning resources to the RIR when they are not being used anymore. So, if there is a plan to "transfer" those resources to a (specific) 3rd party, it probably means they are not being used, so they should firstly be returned to our RIR... if current rules have any value. The notion/definition that numbering resources are not an asset is present in several RIPE documents, correct? if this proposal is approved, imho some work needs to be done urgently to "ammend/remove" the documentation phrases where that definition is expressed/implied... Regards, Carlos Friaças On Fri, 13 Apr 2012, Gert Doering wrote:
Hi,
On Fri, Apr 13, 2012 at 06:57:16AM +0000, Tero Toikkanen wrote:
Also, we already have a policy for transferring address space within the RIPE-region.
"within", yes.
Why is this proposal trying to create a new policy instead of amending what we already have in place?
Because our current policy doesn't say anything about transfers outside of the RIPE region (plus, our current policy also needs work, but that's a different thread).
Out-of-region transfers are (obviously) different because the rules of the corresponding RIR need to be taken into account.
Gert Doering -- APWG chair -- have you enabled IPv6 on something today...?
SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
On 13 Apr 2012, at 09:52, Carlos Friacas wrote:
I may be wrong, but i guess this community has some rules about returning resources to the RIR when they are not being used anymore. So, if there is a plan to "transfer" those resources to a (specific) 3rd party, it probably means they are not being used, so they should firstly be returned to our RIR... if current rules have any value.
Carlos, this is true in an abstract, academic sense. However it won't apply in the real world. For one thing, it doesn't necessarily follow that those addresses are not used: the donor (seller) just has no need for them any more. A "business reorganisation" could mean those addresses and the customers using them move to a new entity. Transfers between LIRs can already be gamed within the existing policies. Expect to see more of this once v4 has run out. So please, let's not kid ourselves that blocks of "unused" IPv4 addresses will always be returned to the RIR who issued them. That will be even less likely when those blocks are perceived to have significant value once the RIRs have no more v4 resources. IMO it's far more important that the RIRs keep track of who's got which blocks of addresses and maintain an accurate, timely database of that info. An address policy which reflects that will be better than one which doesn't.
On 04/13/2012 02:03 PM, Jim Reid wrote:
Transfers between LIRs can already be gamed within the existing policies.
imho no. - ips cannot be owned - allocation is tied to need - transfers are treated identical to allocations => transfers between lirs are subject to the same prerequisites as allocations. transfers can be gamed no more or less than allocations. this is the formal situation. if sbdy is trying to or letting sbdy get through with some sort of cheat, that's a totally different problem... regards, Chris
On 13 Apr 2012, at 13:19, Chris wrote:
transfers between lirs are subject to the same prerequisites as allocations.
Yes. I have read the policy. However there are transfers and there are transfers. A formal transfer according to the policy is of course bound by those needs-based prerequisites. Address blocks sometimes need to be moved between LIRs without invoking that: no back-room deals or cheating either. You just don't call those moves a transfer, even though that's the end result. Suppose a hypothetical IXP sells off its web hosting business. It keeps some IP addresses for itself and the rest go with the web hosting. The web customers can't be renumbered but they move to another company which might or might not be an existing LIR in the same service region. What sort of "needs-based address transfer" is that? This is a rhetorical question BTW.
On 04/13/2012 02:36 PM, Jim Reid wrote:
Yes. I have read the policy. However there are transfers and there are transfers. A formal transfer according to the policy is of course bound by those needs-based prerequisites. Address blocks sometimes need to be moved between LIRs without invoking that: no back-room deals or cheating either. You just don't call those moves a transfer, even though that's the end result.
i'm convinced we both agree that we are not talking about simple announcements of one lir's pa by another lir's as. the rest, which i think you are actually referring to, is very well known to me, too. and i can tell you the examples coming to my mind are totally solvable by sticking to the policy, you just have to mean it (e.g. i personally never bothered as long as ip space wasn't really close to depletion). some closer research into this even showed that by just dealing with the very worst cases, the 'exhausted'-state would just vanish - and chances are, for a long time (i guess at least 5 years. i think if played right, actually it had the potential to kind of solve the 'hard' depletion issue entirely, while at the same time pushing v6).
Suppose a hypothetical IXP
you just HAD to pick IXPs here, hm? ;)
sells off its web hosting business. It keeps some IP addresses for itself and the rest go with the web hosting. The web customers can't be renumbered but they move to another company which might or might not be an existing LIR in the same service region. What sort of "needs-based address transfer" is that? This is a rhetorical question BTW.
too bad, because i'll answer that anyway: first of all, just a little reminder: one of the first things every lir is informed about is that it might have to renumber in some cases. period. the ixp sells off its web hosting business: -> it stays lir -> its originally demonstrated need for ips isn't valid anymore -> its right towards ip space is reduced to what its allocation would be according to its current need another lir buys the webhosting business: -> it doesn't buy any ip-addresses -> its needs for ip space are now probably higher than the need demonstrated for its old allocation -> in this case it is eligible for further allocation - probably quite the size of what the ixp selling the web hosting business 'lost' -> from my experience i'd assume the folks at ncc have no problem at all to handle the allocation-recovery and allocation-enlargement in the least-invasive way, i.e. moving the space returned by the ixp to the webhoster with maximum congruity as aggregation allows -> in case an additional block is needed to be allocated to the webhoster (which is rather unlikely but possible), this is also done. in this case the webhoster would have to renumber a small part of the webhosting-farm. all this might be done not by returning/reallocation, but by transfer. same thing, less "ripe" strings in the text. and that's simply the way it works if you'd stick to the current situation, and _don't_ cheat or look the other way sometimes or change the fundamental policy of "ips cannot be owned". => the most interesting thing here is: the ip situation is identical to the "ok now we regard ips as asset and let people sell and buy them" case. if the space is sold, we have the same - identical - problem, we have to deal with aggregation and possibly renumbering in the same way. the only difference is that theoretically someone could simply buy up all the ip space he wants, maybe just letting it rot or abusing it in whatever way, and others had no chance to get any. regards, Chris
Because our current policy doesn't say anything about transfers outside of the RIPE region (plus, our current policy also needs work, but that's a different thread).
Out-of-region transfers are (obviously) different because the rules of the corresponding RIR need to be taken into account.
I was merely questioning the rationale of having a completely separate transfer policy for each case, in stead of just one covering all cases. In that case we could have general conditions covering all cases and more detailed provisions for individual transfer types. Right now the conditions presented in the proposal seem to be lacking in many perspectives, as has already been noted on this discussion. I also know that the current intra-RIR transfer policy is embedded in the more general allocation and assignment policy, so in that sense it seems like a good idea to make a separate policy and not tamper with existing material. Should the community feel that a policy for inter-RIR transfers is needed, I urge to keep policy coherency in mind when considering conditions. Right now I feel that the suggested conditions for inter-RIR transfers differ too much from conditions already in place for intra-RIR transfers. For example, in case of intra-RIR transfers, the need for address space is evaluated by RIPE, which (according to the proposal) would not be the case with inter-RIR transfers (the wording of condition 2. is quite vague). ____________________________________ Tero Toikkanen Nebula Oy
Hi, On Fri, Apr 13, 2012 at 09:14:55AM +0000, Tero Toikkanen wrote:
I was merely questioning the rationale of having a completely separate transfer policy for each case, in stead of just one covering all cases. In that case we could have general conditions covering all cases and more detailed provisions for individual transfer types. Right now the conditions presented in the proposal seem to be lacking in many perspectives, as has already been noted on this discussion.
Yes, this is a good point, and a little bird told me that we'll see something to address this presented next week at the AP meeting :-) Gert Doering -- APWG chair -- have you enabled IPv6 on something today...? SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
-----Original Message----- I was merely questioning the rationale of having a completely separate transfer policy for each case, instead of just one covering all cases. [Milton L Mueller]
The problem is that you have 5 different RIRs and (potentially) one global market.
In that case we could have general conditions covering all cases and
[Milton L Mueller] which "we" are you talking about? The European "we", American we....?
I also know that the current intra-RIR transfer policy is embedded in the more general allocation and assignment policy, so in that sense it seems like a good idea to make a separate policy and not tamper with existing material. Should the community feel that a policy for inter-RIR transfers is needed, I urge to keep policy coherency in mind when considering conditions. Right now I feel that the suggested conditions for inter-RIR transfers differ too much from conditions already in place for intra-RIR transfers. For example, in case of intra-RIR transfers,
[Milton L Mueller] But that is precisely the crux of the problem we are facing. If you think it's hard to get a single coherent policy through one RIR/state, trying getting one through 5, especially when the incentives of the members of the different territorial communities diverge. Territorially exclusive RIRs are very similar in structure and incentives to territorially sovereign states, and defining a single, uniform policy for inter-RIR transfers is very much like negotiating a trade treaty. In short, I think a global policy should completely supplant the regional/territorial policy.
the need for address space is evaluated by RIPE, which (according to the proposal) would not be the case with inter-RIR transfers (the wording of condition 2. is quite vague).
[Milton L Mueller] If you think the global policy should try to conform to the local ones, you've got it backwards. It is the opposite that needs to occur.
On 14 April 2012 00:52, Milton L Mueller <mueller@syr.edu> wrote:
[Milton L Mueller] But that is precisely the crux of the problem we are facing. If you think it's hard to get a single coherent policy through one RIR/state, trying getting one through 5, especially when the incentives of the members of the different territorial communities diverge. Territorially exclusive RIRs are very similar in structure and incentives to territorially sovereign states, and defining a single, uniform policy for inter-RIR transfers is very much like negotiating a trade treaty.
In short, I think a global policy should completely supplant the regional/territorial policy.
... but what required is not a single policy globally but rather 5 (actually you would technically need more for the NIR that exist) policies that interoperate. How they do this depends on the RIRs to decide for themselves: 1. Afrinic - "We do not support inter RIR transfers" 2. ARIN - "We support transfers to RIRs where there is a policy that exists to support the reverse" 3. RIPE - "We've not decided so until then they are not supported" etc All these policies are interoperable, its only where they don't mesh that there is a problem. RIPE needs a policy (even if the policy is "we don't support this") and then other regions can get their policy to match up, I hope that the policy that is about to be released to this list (with other authors) can solve this problem. J -- James Blessing 07989 039 476
On Fri, Apr 13, 2012 at 11:52:29PM +0000, Milton L Mueller wrote:
-----Original Message----- I was merely questioning the rationale of having a completely separate transfer policy for each case, instead of just one covering all cases. [Milton L Mueller]
The problem is that you have 5 different RIRs and (potentially) one global market.
kind of like spectrum. /bill
Hi, On Sat, Apr 14, 2012 at 11:10:14AM +0000, bmanning@vacation.karoshi.com wrote:
The problem is that you have 5 different RIRs and (potentially) one global market.
kind of like spectrum.
Not at all, given that you can't just use a spectrum licensed in the US in Europe. Gert Doering -- APWG chair -- have you enabled IPv6 on something today...? SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
On Sat, Apr 14, 2012 at 02:38:26PM +0200, Gert Doering wrote:
Hi,
On Sat, Apr 14, 2012 at 11:10:14AM +0000, bmanning@vacation.karoshi.com wrote:
The problem is that you have 5 different RIRs and (potentially) one global market.
kind of like spectrum.
Not at all, given that you can't just use a spectrum licensed in the US in Europe.
actually, in some cases you can. (my radio works just fine in Europe) in other cases, its just like NAT.. :) if you can spare some cycles, its worth a look at the World Radio Congress. Global coordination of a "commons" that is done on regional and national basis. They have solved lots of the problems the RIRs are contemplating now. That said, its not identical but the parallels are many and close. Twould be a shame to ignore their work just because some take an NIH view of things... NIH - Not Invented Here. /bill
Gert Doering -- APWG chair -- have you enabled IPv6 on something today...?
SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
[spectrum regulation]
They have solved lots of the problems the RIRs are contemplating now.
I'm not sure if it's instructive. There are a few odd things about spectrum allocations: There's no global uniqueness requirement. Not even the frequency ranges for the next-layer protocols are standardized globally. You have to throw away devices when the allocation changes (or at least disable the RF interface). With such possibilities at hand, the solution space is somewhat different.
On 14 Apr 2012, at 00:52, Milton L Mueller wrote:
In short, I think a global policy should completely supplant the regional/territorial policy.
Feel free to propose such a policy. This would be much more productive than discussing academic theories on how best to deal with this issue. For some definition of "best". Where and who would develop this global policy?
participants (10)
-
bmanning＠vacation.karoshi.com -
Carlos Friacas -
Chris -
Erik Bais -
Florian Weimer -
Gert Doering -
James Blessing -
Jim Reid -
Milton L Mueller -
Tero Toikkanen