On Fri, May 20, 2016 at 3:17 PM, Gert Doering <gert@space.net> wrote:

Dear Working Group,

On Fri, May 20, 2016 at 03:02:43PM +0200, Marco Schmidt wrote:

...

The Discussion Period for the proposal 2015-05, "Last /8 Allocation Criteria Revision" has been extended until 13 June 2016.

this has been decided by proposers and WG chairs based on your discussion and the upcoming AP meeting at RIPE72 (next wednesday) - keep the proposal active until after the discussion there (see below), then decide how to proceed.

From the discussion it was very clear that there is no consensus today to go ahead - without going into detail, it's clear that there are two strong factions, one that wants to preserve the remaining /22s for "as long as possible", while the other one wants to ease the pain for those LIRs that have too little IPv4 today, willing to incur earlier total run-out as a consequence.

Since we've supposed to work toward something that can gain consensus I've got a few questions for the authors, and those supporting 2015-05. So far all I've heard, I might have missed something, is that there is a need for more addresses. None have said why, or where there is a need. Why do you need more addresses and for what? Be specific, is it for having more address for the end-users? Datacenter? Services? Infrastructure? IPv6-to-IPv4 services? CGN? Proxyes? -- Roger Jorgensen | ROJO9-RIPE rogerj@gmail.com | - IPv6 is The Key! http://www.jorgensen.no | roger@jorgensen.no

Hi, On Sat, May 21, 2016 at 11:05:16AM +0100, Nick Hilliard wrote:

Roger Jørgensen wrote:

...

Be specific, is it for having more address for the end-users? Datacenter? Services? Infrastructure? IPv6-to-IPv4 services? CGN? Proxyes?

[x] all of the above, and more.

This question isn't relevant as it seems - lots of organisations have their needs and the RIPE NCC cannot and should not be arbiter of whose need is greatest or should take precedence.

Right - *but* it might be an interesting idea to turn around this discussion, away from haggling about the last scraps, into being able to give more useful guidance to LIRs. Like, - if you need to connect end-users, best practice is dual-stack with native IPv6 and CGNAT IPv4 (it stinks, but gets the job done while content is not IPv6 capable everyhwere) - if you run a data-center, run ipv6-only on the inside, and add Tore-style NAT46 to give each service a single public IPv4 address (insert pointer to RFC...) etc. While not truly *APWG* relevant, we could at least find out where the highest pain is, and then throw the ball over to the IPv6 WG to provide solutions :-) (totally IETF style). Gert -- 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 (0)89/32356-444 USt-IdNr.: DE813185279

Hi,

Op 21 mei 2016, om 17:52 heeft Nick Hilliard <nick@foobar.org> het volgende geschreven:

Gert Doering wrote:

...

Right - *but* it might be an interesting idea to turn around this discussion, away from haggling about the last scraps, into being able to give more useful guidance to LIRs.

Like,

- if you need to connect end-users, best practice is dual-stack with native IPv6 and CGNAT IPv4 (it stinks, but gets the job done while content is not IPv6 capable everyhwere)

- if you run a data-center, run ipv6-only on the inside, and add Tore-style NAT46 to give each service a single public IPv4 address (insert pointer to RFC...)

etc.

While not truly *APWG* relevant, we could at least find out where the highest pain is, and then throw the ball over to the IPv6 WG to provide solutions :-) (totally IETF style).

That sounds like an offer to write the document.

Hans Petter just made a nice statement in the opening plenary. I'm taking this to the BCOP session this afternoon to see if we can write this down properly. Cheers, Sander

Hi, On Sun, May 22, 2016 at 10:29:37AM +0200, Riccardo Gori wrote:

every policy that makes IPv6 adoption a must can help slow down IPv4 allocation rate and in the meanwhile will even lower IPv4 maket value that's why there is the so called "no solution" sorry for brevity, family time....

The problem with that is: by imposing "YOU MUST DO IPv6!" on requestors that only have a /22 yet (assuming that we could find a meaningful way to check IPv6 deployment that cannot be cheated) - you're not going to solve the thing that people are complaining about in this thread: "old LIRs that have enough IPv4 space and are not deploying IPv6!" New LIRs - holders of /22 - have all the incentives to deploy IPv6 already (because they do not have enough IPv4 to number everything with public v4 addresses) - but how would such a policy incentivize a big content provider that has enough v4, is not growing in number of external visible services (= doesn't need more v4 addresses), and has no v6? These are the sore spots today: content and cloud providers - and neither are likely to fall under this policy. 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 (0)89/32356-444 USt-IdNr.: DE813185279

Riccardo Gori wrote:

...

There are no good solutions to the problem at hand, only compromises. If the current policy is changed to something else, the people who benefit in the short term will be happier and the people who pay for this generosity will be disappointed. IPv6

If ipv6 provided full backwards compatibility with ipv4, it would be a solution. Any organisation which is forced into providing ipv4 connectivity using an ipv6 transition mechanism will operate at a competitive disadvantage to an organisation which can provide native ipv4. This isn't a swipe at ipv6: it's simply an observation that we have come to depend on features of ipv4 which cannot be fully replicated using ipv6 transition mechanisms. This is why I referred to migrating to ipv6 as a compromise. Long term, it may be a good compromise and many years down the road, it may even be better than ipv4. But as long as there are ipv4 addresses available, it is condemned to being second best by the legacy of the existing single-stacked ipv4 install base.

every policy that makes IPv6 adoption a must can help slow down IPv4 allocation rate and in the meanwhile will even lower IPv4 maket value [...]

I don't believe there is any evidence to support either of these statements. Also, please bear in mind that if strong policies caused better adoption of new protocols, we would be having this discussion using X.400. Nick

On Fri, May 20, 2016 at 3:17 PM, Gert Doering <gert@space.net> wrote:

...

Dear Working Group,

...

The Discussion Period for the proposal 2015-05, "Last /8 Allocation Criteria Revision" has been extended until 13 June 2016.

On Fri, May 20, 2016 at 03:02:43PM +0200, Marco Schmidt wrote: this has been decided by proposers and WG chairs based on your discussion and the upcoming AP meeting at RIPE72 (next wednesday) - keep the proposal active until after the discussion there (see below), then decide how to proceed.

From the discussion it was very clear that there is no consensus today to go ahead - without going into detail, it's clear that there are two strong factions, one that wants to preserve the remaining /22s for "as long as possible", while the other one wants to ease the pain for those LIRs that have too little IPv4 today, willing to incur earlier total run-out as a consequence. Since we've supposed to work toward something that can gain consensus I've got a few questions for the authors, and those supporting 2015-05.

So far all I've heard, I might have missed something, is that there is a need for more addresses. None have said why, or where there is a need. Why do you need more addresses and for what? In my opinion there's a trend change from what happened years before. Standing on current allocation rate and LIRs sign up rate we can see

Hi Roger, thank you for your questions. I try to answer below Il 21/05/2016 09:45, Roger Jørgensen ha scritto: there's general trend in the internet growth. Internet grew a lot up to 2000-2002 then the trend was pretty the same up to few year ago. I think it's mainly due to the fact that in many countries become an "operator" is easier than before As example in Italy we had many restrictions that made things very difficoult without a good capital fund behind. Nowadays things are easier and many new small operators are coming to the market. The WISP market grew a lot in these years look at You can look at Ubiquity or Candium Networks sells. In Italy is the fiber time and carry and transport fiber optics is easier than before. So, in my opinion that's why we are registering such big growth trend in LIR sign ups. In the meanwhile Last /8 allocation policy in showing out its limits that created some competitive problems to new entrants. And at the same time we (as community) and RIPE NCC as our point of coordination weren't able to provide a reable solution to help IPv6 adoption. Small operators has less capitals, rsources and experience but the policy asked them to act before the big ones. 2015-05 trys to address competivive disvantage of new entrants and small operators while reminding everyone there's only a solution: IPv6. If you try to adopt it you may find your way out of the problem. The clear suspect I have is that there are too many interest to keep calm about IPv6 to save market value of IPv4 as long as possibile.

Be specific, is it for having more address for the end-users? Datacenter? Services? Infrastructure? IPv6-to-IPv4 services? CGN? Proxyes?

It's happening: end customers of new operators (read as new LIRs) are requesting new services such as datacenters or multihoming and IPv6 deployment in the meanwhile. Those are the tipical request that I reiceve. For example to multihome and bgp a customer I need a /24 What if I have no address space to provide? I can ask my customer to sign up and he will get a /22 automatically wasting a 3 x /24 I think in many cases this is why we are registering such new sign up growth trends. I already said in past emails that when I started our business of fiber optic provider the carrier said to us "ask us for transport and access but not for addresses. sign up and get yours" This is reflecting in all the chain from top to bottom. This could be a point where to act. If we turn the request re-introducing justification and we turn minimum request to a /24 we can address this kind of problem while slowing down LIRs sign up rate to obtain a /23 or /24 to address this kind of requests hope this help in understand small player needings regards Riccardo -- Ing. Riccardo Gori e-mail: rgori@wirem.net Mobile: +39 339 8925947 Mobile: +34 602 009 437 Profile: https://it.linkedin.com/in/riccardo-gori-74201943 WIREM Fiber Revolution Net-IT s.r.l. Via Cesare Montanari, 2 47521 Cesena (FC) Tel +39 0547 1955485 Fax +39 0547 1950285 -------------------------------------------------------------------- CONFIDENTIALITY NOTICE This message and its attachments are addressed solely to the persons above and may contain confidential information. If you have received the message in error, be informed that any use of the content hereof is prohibited. Please return it immediately to the sender and delete the message. Should you have any questions, please contact us by re- plying to info@wirem.net Thank you WIREM - Net-IT s.r.l.Via Cesare Montanari, 2 - 47521 Cesena (FC) --------------------------------------------------------------------

Hi, Am 22.05.2016 um 12:01 schrieb Tore Anderson:

* Riccardo Gori

...

and we turn minimum request to a /24 we can address this kind of problem while slowing down LIRs sign up rate to obtain a /23 or /24 to address this kind of requests This, in isolation, I think is idea worth exploring further.

The minimum allocation size started out as a /19 (cf. ripe-136). Over time it's been adjusted down to the current /22. I think it might be prudent to continue this trend at some point.

For example: change it to a /23 at the point when the NCC pool reaches the equivalent of a /9, and then to a /24 when the pool reaches the equivalent of a /24.

This would ensure the last /9-equivalent could accomodate three times as many new entrants (24576) than if we continue with /22s (8192).

One would hope that even though the future new entrants will continue to need some IPv4 to start their business, the amount will decrease as IPv6 deployment increases. That is, a new entrant receiving a final /23 in 2017 might find it about as useful the final /22 was to a new entrant in 2013.

Depending on how the membership fees and the pricing in the second-hand IPv4 market develops, this could also help discourage abuse: if everything else remains the same, the cost per address would essentially double at each adjustment.

Another thing worth considering (separately) is to stop issuing additional allocations. An "old LIR" (one that held IPv4 allocations on the 14th of September 2012 that hasn't needed to request its final /22 in the four years since, is likely not growing anyway and that space is better spent for new entrants. Four years is in any case a much longer period than the LIRs last allocation was supposed to last.

Basically we'd need to change bullet #2 in ripe-649 section 5.1 to just say «an LIR that currently hold or have previously held an IPv4 allocation is not eligible» or something like that. As an added bonus, we'd then get rid of the quirky unexplained 14-09-2012 date referenced in the current text.

This would prevent "old LIRs" that are already holding lots of address space, perhaps mostly unused, from being able to pick up a final /22 anyway. I can easily see that a "new LIR" making very efficient use of its final /22 while being unable to request another would find this very unfair, as would the new entrants that inevitably would be denied any final IPv4 allocations down the line (due to full depletion happening earlier).

Unfortunately I won't be in Copenhagen so just consider this me thinking out loud here on the list instead of at the Open Policy Hour. I'm not going to submit any proposals myself though so anyone should feel free to take the above ideas and run with them.

Tore

in a membership organisation we have to treat all members equal when it comes to distributing the final IPv4 networks. You cannot take the right to request the final /22 away just because a member had no need so far to ask for it. This should have been put in the policy a few years ago (like "you have 4 years to request your /22") , but you cannot change this today working backward. Wilhelm

* Gert Doering

OTOH, I'm not sure if I see a pressing need here - LIRs that haven't asked for their /22 yet because they don't need it might just never show up, because they don't need it...

Absolutely true. However, in my opition it would be more about fixing a couple of perception problems that leads to policy proposals like 2015-05: - Complaints à la «it's unfair that the haves ("old LIRs") get to request additional allocations while we have-nots ("new LIRs") do not» will no longer have any merit. - It'd make sure the NCC's allocation pool is really reserved for new entrants. We often claim that it is, but currently that is only a half-truth. Tore

two people said i was not sufficiently obvious, so i hacked a little

* Riccardo Gori

Hi again Tore,

Il 22/05/2016 12:01, Tore Anderson ha scritto:

...

* Riccardo Gori

...

and we turn minimum request to a /24 we can address this kind of problem while slowing down LIRs sign up rate to obtain a /23 or /24 to address this kind of requests This, in isolation, I think is idea worth exploring further.

The minimum allocation size started out as a /19 (cf. ripe-136). Over time it's been adjusted down to the current /22. I think it might be prudent to continue this trend at some point.

For example: change it to a /23 at the point when the NCC pool reaches the equivalent of a /9, and then to a /24 when the pool reaches the equivalent of a /24.

This would ensure the last /9-equivalent could accomodate three times as many new entrants (24576) than if we continue with /22s (8192).

One would hope that even though the future new entrants will continue to need some IPv4 to start their business, the amount will decrease as IPv6 deployment increases. That is, a new entrant receiving a final /23 in 2017 might find it about as useful the final /22 was to a new entrant in 2013.

Depending on how the membership fees and the pricing in the second-hand IPv4 market develops, this could also help discourage abuse: if everything else remains the same, the cost per address would essentially double at each adjustment.

Another thing worth considering (separately) is to stop issuing additional allocations. An "old LIR" (one that held IPv4 allocations on the 14th of September 2012 that hasn't needed to request its final /22 in the four years since, is likely not growing anyway and that space is better spent for new entrants. Four years is in any case a much longer period than the LIRs last allocation was supposed to last.

Basically we'd need to change bullet #2 in ripe-649 section 5.1 to just say «an LIR that currently hold or have previously held an IPv4 allocation is not eligible» or something like that. As an added bonus, we'd then get rid of the quirky unexplained 14-09-2012 date referenced in the current text.

This would prevent "old LIRs" that are already holding lots of address space, perhaps mostly unused, from being able to pick up a final /22 anyway. I can easily see that a "new LIR" making very efficient use of its final /22 while being unable to request another would find this very unfair, as would the new entrants that inevitably would be denied any final IPv4 allocations down the line (due to full depletion happening earlier).

Unfortunately I won't be in Copenhagen so just consider this me thinking out loud here on the list instead of at the Open Policy Hour. I'm not going to submit any proposals myself though so anyone should feel free to take the above ideas and run with them.

those above statements are the most constructives. So we agree toghether that last /8 had its good effects but can be tuned to adapt itself to the current times coming?

I believe the «last /8» policy so has been, and continue to, work very well. If not for that policy, the new members joining since 2013 or so would not have been able to request any IPv4 at all. Of course, it is certainly possible to tune it further - after all, the RIPE community has done precisely that on several occasions already. I don't see it as a *necessity* though; that is, the policy isn't broken (it's working as intended and quite well too) so it doesn't really need fixing. I'd be happy enough if we leave it as-is too. Tore

Hi Roger, Il 23/05/2016 14:38, Roger Jørgensen ha scritto:

On Sun, May 22, 2016 at 10:25 AM, Riccardo Gori <rgori@wirem.net <mailto:rgori@wirem.net>> wrote:

Hi Roger,

thank you for your questions. I try to answer below

Il 21/05/2016 09:45, Roger Jørgensen ha scritto:

...

<snip>

...

Be specific, is it for having more address for the end-users? Datacenter? Services? Infrastructure? IPv6-to-IPv4 services? CGN? Proxyes?

It's happening: end customers of new operators (read as new LIRs) are requesting new services such as datacenters or multihoming and IPv6 deployment in the meanwhile. Those are the tipical request that I reiceve. For example to multihome and bgp a customer I need a /24 What if I have no address space to provide? I can ask my customer to sign up and he will get a /22 automatically wasting a 3 x /24 I think in many cases this is why we are registering such new sign up growth trends. I already said in past emails that when I started our business of fiber optic provider the carrier said to us "ask us for transport and access but not for addresses. sign up and get yours" This is reflecting in all the chain from top to bottom. This could be a point where to act. If we turn the request re-introducing justification and we turn minimum request to a /24 we can address this kind of problem while slowing down LIRs sign up rate to obtain a /23 or /24 to address this kind of requests

hope this help in understand small player needings

You have given me no real reason, just nice to have. We passed nice to have some years ago.

End users cannot continue to get a /24, there are not enough address space for that, sorry but that's life. Sure some operators have enough and that's unfair for others. Only way for them to get something like that is to either become LIR, or use IPv6. Why is it so hard to understand that? I meant a datacenter to be multihomed. I consider my customers that hosts a datacenter to be multihomed to be end-users since they are interested only in their own business and don't care about IP or LIRs stuff. They just need good quality internet presence and availability. My tipical business customer is software houses / saas clusters hosting company / or datacenter itself. I would use "consumer" for end user home customer.

So I ask _again_, where is the IPv4 need? What type of usage is it ment for? We've passed the "it's nice to have" some years ago, now we're down to , do you _really_ need 10 addresses? Can you survive with 2 and deploy IPv6?

I think I answered, It's not nice to have, It's business demand and LIRs should be able to offer... with a /22 I can serve just up to 2 or 3 of my tipical business customers. This is lack of competitiveness.

--

Roger Jorgensen | ROJO9-RIPE rogerj@gmail.com <mailto:rogerj@gmail.com> | - IPv6 is The Key! http://www.jorgensen.no | roger@jorgensen.no <mailto:roger@jorgensen.no>

regards Riccardo -- Ing. Riccardo Gori e-mail: rgori@wirem.net Mobile: +39 339 8925947 Mobile: +34 602 009 437 Profile: https://it.linkedin.com/in/riccardo-gori-74201943 WIREM Fiber Revolution Net-IT s.r.l. Via Cesare Montanari, 2 47521 Cesena (FC) Tel +39 0547 1955485 Fax +39 0547 1950285 -------------------------------------------------------------------- CONFIDENTIALITY NOTICE This message and its attachments are addressed solely to the persons above and may contain confidential information. If you have received the message in error, be informed that any use of the content hereof is prohibited. Please return it immediately to the sender and delete the message. Should you have any questions, please contact us by re- plying to info@wirem.net Thank you WIREM - Net-IT s.r.l.Via Cesare Montanari, 2 - 47521 Cesena (FC) --------------------------------------------------------------------

Hi Roger, Il 23/05/2016 14:58, Roger Jørgensen ha scritto:

On Mon, May 23, 2016 at 2:48 PM, Riccardo Gori <rgori@wirem.net <mailto:rgori@wirem.net>> wrote:

Hi Roger,

Il 23/05/2016 14:38, Roger Jørgensen ha scritto:

...

On Sun, May 22, 2016 at 10:25 AM, Riccardo Gori <rgori@wirem.net <mailto:rgori@wirem.net>> wrote:

Hi Roger,

thank you for your questions. I try to answer below

Il 21/05/2016 09:45, Roger Jørgensen ha scritto:

...

<snip>

So I ask _again_, where is the IPv4 need? What type of usage is it ment for? We've passed the "it's nice to have" some years ago, now we're down to , do you _really_ need 10 addresses? Can you survive with 2 and deploy IPv6?

I think I answered, It's not nice to have, It's business demand and LIRs should be able to offer... with a /22 I can serve just up to 2 or 3 of my tipical business customers. This is lack of competitiveness.

sorry if I sound hard/insulting, to me it sounds like you have an outdated business model that are doomed to go down soon, you can't grow this way. Change to IPv6 where you actual can have some growth and option for the future.

What you are illustrating is that somehow, despite all the effort from many parties the world do not seems to understand that there are no more IPv4, the growth has to come from IPv6. if you look at the policy proposal we are just addressing the same point. You'll get some more if your are aware that IPv6 is the solutions and you demonstrate you are doing something.

--

Roger Jorgensen | ROJO9-RIPE rogerj@gmail.com <mailto:rogerj@gmail.com> | - IPv6 is The Key! http://www.jorgensen.no | roger@jorgensen.no <mailto:roger@jorgensen.no> regards Riccardo --

Ing. Riccardo Gori e-mail: rgori@wirem.net Mobile: +39 339 8925947 Mobile: +34 602 009 437 Profile: https://it.linkedin.com/in/riccardo-gori-74201943 WIREM Fiber Revolution Net-IT s.r.l. Via Cesare Montanari, 2 47521 Cesena (FC) Tel +39 0547 1955485 Fax +39 0547 1950285 -------------------------------------------------------------------- CONFIDENTIALITY NOTICE This message and its attachments are addressed solely to the persons above and may contain confidential information. If you have received the message in error, be informed that any use of the content hereof is prohibited. Please return it immediately to the sender and delete the message. Should you have any questions, please contact us by re- plying to info@wirem.net Thank you WIREM - Net-IT s.r.l.Via Cesare Montanari, 2 - 47521 Cesena (FC) --------------------------------------------------------------------

* Riccardo Gori

I think I answered, It's not nice to have, It's business demand and LIRs should be able to offer... with a /22 I can serve just up to 2 or 3 of my tipical business customers. This is lack of competitiveness.

So, let me get this straight: In order to facilitate growing your business beyond three customers, you've submitted a RIPE policy proposal that'd let you get eventually another 3*/22 from the free pool. This would then allow you to grow your business to having 8-12 customers. Right? Assuming 2015-05 does go through: when your prospective customer number 13 is knocking on your door, should we then expect you to return with another policy proposal to change 2015-05 /20 ceiling to /18? Assuming that proposal also goes through: should we expect you to return again, asking the RIPE community to extend the /18 ceiling by another two bits, so that you can take on customer number 49? And so on... In any case, it is inevitable that at some point in time the RIPE NCC will simply not have any IPv4 address space to give you, regardless of what the policy allows. What will you do then, exactly? And why aren't you already doing it today? Tore

Hi Ricardo,

If your read again 2015-05 you can easily find out that is not so silly. Currently the only reasonable objection about 2015-05 is that may (I underline may) speed up the allocation rate. Please note that this ojection is based on the insinuation [...]

Ok, this is enough. This is not an insinuation. There have been plenty of numbers that show the effects that 2015-05 will have. They have been discussed, and there is clearly no consensus. Let's stop this right here. It is clear that 2015-05 doesn't have consensus and never will get consensus. I am sorry, you have done your best to get this policy proposal through the process, but you didn't succeed. We are going to discuss which direction to move in on Thursday at the RIPE meeting. We will report to this list based on that. But don't start saying that the objections to your policy proposal are insinuations... Cheers, Sander

Hi Ricardo,

I am not expecting consensum. I just want to be serius about the considerations and the discussion that can come to some costructive for everyone.

We have had that discussion here on the list. Let's finish this with a constructive discussion on Thursday.

Someone choosed to turn it into joke and irony.

After talking about a policy proposal that is not leading to consensus for multiple discussion periods, I can't blame them. People seem to want to close this discussion. Let's do so on Thursday and discuss where to go from there. Cheers, Sander

On Mon, May 23, 2016 at 9:43 PM, Riccardo Gori <rgori@wirem.net> wrote:

<snip> Rorger asked to explain what kind of needs or problems the policy is going to address and I reported clear examples. Current LIRs sign up rate is affacted by LIRs asking their end-users to sign up to save their ip resources and these customers are wasting space creating other unused space you can find this in 2015-05 and much more

you gave me no good answer except demonstrated that you haven't yet realized that there is no more IPv4 to get. -- Roger Jorgensen | ROJO9-RIPE rogerj@gmail.com | - IPv6 is The Key! http://www.jorgensen.no | roger@jorgensen.no

Hi *, initially reading, nowadays just browsing over the posts about 2015-05, I don't get it ... Yes, I understand the pain new LIRs have with the limited availability of v4 addresses. We have a small local provider, a small city that has gone through the trouble of installing a FO infrastructure to provide high-speed internet to most of its citizens, and has to do with their /22 assignment (they became an LIR to get those) for something around (currently) 3k customers, of those something like 100+ businesses. They run native v6 dual-stacked with CGN v4, which - especially for the business customers - isn't the optimal solution (just using a /29 for each business would more or less deplete their public v4 range). So yes, they would love to get more v4, but they can't. I keep reading phrases like "it's not fair", "it's anti-competitive". Hell yes it is. But business has never been fair. If you're coming late to a supper, you may not get all the best pieces of the food that was available at the beginning. If you're late to the business, many customers will already have found a different provider for their requirements and will often not even talk to you. So what, you have to live with it. You knew you were a late entry, if you didn't know what you were getting into, if you are surprised by the limitations of both the market itself and RIPE specifically, you obviously didn't do your homework! Heck, even if you already have a /20, /19 or whatever and are running out of v4 addresses - why are you surprised? v4 was running out a long time ago, which was only delayed by NAT "technology", a kluge to allow for a better technology (IPv6) to be finished and rolled out - 15 years later and too many providers still do not see the necessity of rolling out v6. Who is to blame? Well, of course those providers, most likely ones that still have sufficient amounts of free v4 addresses. To a minor degree possibly their customers, as they ought to be requesting dual-stacked internet from their ISPs. But why should they? Everything is working, and setting up v6 doesn't (in most cases) earn them any benefits. So what could be done? Technically, not much I guess. Sure, we could level the playing field, the RIRs could pass a policy retracting half of all assigned v4 addresses annually from anybody holding - say - /20 or larger. Even going down to 50% would most likely require many LIRs to actually roll out v6, apart from annoy a whole lot of people, as the ISPs would be forced to renumber the remaining IP assignments to their customers. Would such a policy change be enforceable? I doubt it, or at the very least many of the larger ISPs will fight it with all they have (read: law suits). So, as it is, we have a situation in which many large ISPs give a f*ck about the IPv4 scarcity, stick with their pool of v4 addresses and wait until actual customers request v6, which won't happen, so there is no business case to implement it for quite some time. Many other ISPs have to make do, and are at a weaker business position than should be necessary, but there's nothing we as a RIPE community can do except keep up the current policy to at least allow for SOME possibility of getting on the net for new entries, and keep pushing v6 in the market, in a hope to make customers aware of the necessity of v6 to THEIR business! Maybe then more ISPs will finally invest in the future of their part of the Internet. Oh, and just in case you were still wondering: I'm against loosening the last /8 policy ... Garry --

Roger Jørgensen wrote:

We've passed the "it's nice to have" some years ago, now we're down to , do you _really_ need 10 addresses? Can you survive with 2 and deploy IPv6?

No really, we're not. The RIPE Community exited the "do you really need X addresses?" stage in 2014. We're now at the stage where LIRs need to do what they feel is necessary to preserve their remaining stocks of IP addresses in the way that they feel is most appropriate for their long term requirements. There are training courses, policy documents, endless discussions, ARC checks and more advice available than you can shake a stick at, all of which is aimed at helping inform LIRs about depletion and what to do. But ultimately it is up to the LIRs. Neither the RIPE Community nor the RIPE NCC has any business beating them with sticks to get them to plan for their futures. If they do not do this, it is not the responsibility of the RIPE NCC or the RIPE Community to do it for them. If LIRs need more IPv4 address space after their single allocation from the RIPE NCC, then they need to look on the open market. If/when the open market dries up or becomes too expensive, then there will be no options left except ipv6 and/or even more NAT. Nick