On Mon, Sep 10, 2012 at 03:10:46PM +0200, Gert Doering wrote:

...

For this reason alone, I see no major issue with also assigning PI from that final /8...

Is that just a remark, or are you speaking in favour of 2012-04?

Both. Sorry, should have clarified that. rgds, Sascha Luck

+1 -----Original Message----- From: address-policy-wg-bounces@ripe.net [mailto:address-policy-wg-bounces@ripe.net] On Behalf Of Randy Bush Sent: Monday, September 10, 2012 3:55 PM To: Erik Bais Cc: Nick Hilliard (nick@inex.ie); address-policy-wg@ripe.net Subject: Re: [address-policy-wg] [policy-announce] 2012-04 New Draft Document Published (PI Assignments from the last /8)

But I'm afraid that there will be a run on the last possible addresses and that there won't be any IP space left for new LIR's if they require it.

i can not speak to the intent in the ripe region. but the principal intent of the final /8 policy in the apnic region was so that *new* entrants could get one small piece of the pie. in the years when we have a mixed 6/4 internet, folk will need a wee bit of v4 space for things such as nat64 and other ways face the dual stack world. pissing it away in greed and/or arrogance is not responsible to our children. randy Op dit e-mailbericht is een disclaimer van toepassing, welke te vinden is op http://www.esprittelecom.nl/disclaimer/html

Gert, My apoligies - it is a +1 on Randy's email. Lets reserve the last few ipv4 addresses for those who come in the future and not blow through them quickly. So I do not support the change in policy. Jasper -----Original Message----- From: Gert Doering [mailto:gert@space.net] Sent: Monday, September 10, 2012 4:12 PM To: Jasper Jans Cc: Randy Bush; Erik Bais; Nick Hilliard (nick@inex.ie); address-policy-wg@ripe.net Subject: Re: [address-policy-wg] [policy-announce] 2012-04 New Draft Document Published (PI Assignments from the last /8) Hi, On Mon, Sep 10, 2012 at 04:00:36PM +0200, Jasper Jans wrote:

+1

Does that mean "I support the policy proposal 2012-04" or "I agree with Randy about the original intent of the last-/8-Policy" or "I oppose 2012-04"? A *few* more words save lots of extra questions. (I took Randy's e-mail as explanation about the last /8 policy, and not voicing a particular opinion on 2012-04. Correct me if that was not right) 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 Op dit e-mailbericht is een disclaimer van toepassing, welke te vinden is op http://www.esprittelecom.nl/disclaimer/html

Hi Randy,

i can not speak to the intent in the ripe region. but the principal intent of the final /8 policy in the apnic region was so that *new* entrants could get one small piece of the pie. in the years when we have a mixed 6/4 internet, folk will need a wee bit of v4 space for things such as nat64 and other ways face the dual stack world. pissing it away in greed and/or arrogance is not responsible to our children.

I fully agree with you. My company (A2B Internet) did a lot of PI registrations in the last couple of years and it would be very nice to have this available for a couple more months for possible customers in the next couple months. However, having said that, the reserved space in the final /8 is to allow '*new* entrants' and that is also how I view this. So although it might not be in the best interest for my company (PI registration business wise), I still don't agree with the proposal. Regards, Erik Bais

On Mon, Sep 10, 2012 at 04:30:10PM +0200, Sander Steffann wrote:

...

i can not speak to the intent in the ripe region. but the principal intent of the final /8 policy in the apnic region was so that *new* entrants could get one small piece of the pie.

This was also the original reasoning here in the RIPE region.

If so, the letter of the policy contradicts this - there is no mention of only "new" requestors, rather any LIR that can demonstrate a need will be assigned one last /22. One could reasonably assume that the intention has then failed in the original policy and is not a valid argument against the proposal... rgds, Sascha Luck

Hi Sascha,

On Mon, Sep 10, 2012 at 04:30:10PM +0200, Sander Steffann wrote:

...

i can not speak to the intent in the ripe region. but the principal intent of the final /8 policy in the apnic region was so that *new* entrants could get one small piece of the pie.

This was also the original reasoning here in the RIPE region.

If so, the letter of the policy contradicts this - there is no mention of only "new" requestors, rather any LIR that can demonstrate a need will be assigned one last /22. One could reasonably assume that the intention has then failed in the original policy and is not a valid argument against the proposal...

As I understood the policy, is that it isn't only for new LIR's but there is also a reservation for each current LIR. 5.6 Use of last /8 for PA Allocations 1.Allocations for LIRs from the last /8 On application for IPv4 resources LIRs will receive IPv4 addresses according to the following: 1.LIRs may only receive one allocation from this /8. The size of the allocation made under this policy will be exactly one /22. So there is no difference between what a new LIR or a current LIR will receive. As there is only the final /8 soon, new LIR members will get their /22 and that's it. The final /8 does however provide sufficient growth (more than double in numbers of LIR members if I recall correctly) for new LIR members. Regards, Erik Bais

* Randy Bush

i can not speak to the intent in the ripe region. but the principal intent of the final /8 policy in the apnic region was so that *new* entrants could get one small piece of the pie.

in the years when we have a mixed 6/4 internet, folk will need a wee bit of v4 space for things such as nat64 and other ways face the dual stack world.

pissing it away in greed and/or arrogance is not responsible to our children.

Hi Randy, and thanks for bringing up the equivalent APNIC region policy. That is definitively relevant to this discussion. I doubt that allowing organisations to be greedy is the intention of 2012-04, it is rather to also allow non-LIRs to get «a wee bit of v4 space» from the last /8. Currently, this is set aside for LIRs only (both existing and new). Looking at the latest delegated stats file from APNIC, there appears to have been made 510 assignments out of 103/8. They are listed with «ASSIGNED PORTABLE» in whois, which I believe is the equivalent of «ASSIGNED PI» in RIPE land? So it is my understanding that 2012-04 would actually make the RIPE «last /8 policy» more similar to the APNIC one than it currently is. With 2012-04 in place, the largest remaining difference that I could see would be that in the APNIC region assignments up to /22 would be allowed, while in the RIPE region they would be capped at /24 (unlike LIR allocations, which would be capped at /22 - same as APNIC). That said, the «run on the bank» scenario that Erik was concerned about might turn out to be much worse in the RIPE region than in the APNIC region in spite of this. It is my understanding - please correct me if I'm wrong - that in order to get an assignment out of APNIC's last /8, you'll have to become an APNIC member, which currently costs at least AU$1180/year (~€955). If 2012-04 passes, a RIPE region organisation could grab its last /8 piece for €50/year without becoming a RIPE NCC member, which is much less of a barrier. Considering the current pricing and membership structures, getting an last /8 assignment from APNIC is perhaps more comparable to becoming a RIPE region LIR, getting the last /8 allocation, and then make a single assignment to its own infrastructure - which is obviously always an option for an organisation seeking IPv4 address space. -- Tore Anderson Redpill Linpro AS - http://www.redpill-linpro.com/

Hello everyone, This is my first post in any of the RIPE mailing lists, I'm not even sure if I'm qualified or even "allowed". So feel free to put me in place, if needed :) I find Randy's post about the last /8 going to new entrants to be the most convincing one. Giving it away to us (LIRS who are already in possession of at least a /21) should be secondary to giving it to newcomers who really need it in order to get started. I don't mean that's the only group of companies/orgs that should be considered, but they DO need this the most. It's one thing to be tight with the resources you already have, and another to not have them at all. And anyway, how much would life change for an ISP, for example, just cause they got an extra /22? How much time would it really buy? Is that time as valuable as the ability for the new guys to also get SOME ipv4 connectivity? Sorry if I'm out of place, again, I just thought I should try this. Dimitris Ioakimoglou Skype: dioakimoglou Linkedin: http://gr.linkedin.com/pub/dimitris-ioakimoglou/43/510/12 dioakimoglou@globoplc.com | globoplc.com Experience GO!Enterprise Server on your mobile/tablet. Download a demo. -----Original Message----- From: address-policy-wg-bounces@ripe.net [mailto:address-policy-wg-bounces@ripe.net] On Behalf Of Sebastian Wiesinger Sent: Saturday, September 15, 2012 5:58 PM To: address-policy-wg@ripe.net Subject: Re: [address-policy-wg] [policy-announce] 2012-04 New Draft Document Published (PI Assignments from the last /8) * Randy Bush <randy@psg.com> [2012-09-10 16:00]:

i can not speak to the intent in the ripe region. but the principal intent of the final /8 policy in the apnic region was so that *new* entrants could get one small piece of the pie.

in the years when we have a mixed 6/4 internet, folk will need a wee bit of v4 space for things such as nat64 and other ways face the dual stack world.

pissing it away in greed and/or arrogance is not responsible to our children.

I agree with Randy and oppose this proposal. We should not use the last /8 for PI space but for LIRs who need at least a bit of v4 space to get going. Regards Sebastian -- GPG Key: 0x93A0B9CE (F4F6 B1A3 866B 26E9 450A 9D82 58A2 D94A 93A0 B9CE) 'Are you Death?' ... IT'S THE SCYTHE, ISN'T IT? PEOPLE ALWAYS NOTICE THE SCYTHE. -- Terry Pratchett, The Fifth Elephant ________________________________ This e-mail may contain information that is PRIVILEGED and/or CONFIDENTIAL. This e-mail is intended solely for the individual or entity to whom it is addressed. If you are not, or have reason to believe you are not the intended recipient of this e-mail please contact us by reply e-mail as soon as possible and delete the message (including attachment(s)) from your computer system(s) without disclosing it. Any unauthorized review, use, copying, disclosure or distribution of this e-mail (including attachment(s)) is prohibited. The content of this e-mail may contain personal views of its author and does therefore not necessarily reflect the views or policies of Globo (Globo Technologies S.A., Globo Mobile S.A., Profitel S.A, Globo Plc, Globo International LLC ).Globo hereby disclaims any and all liabilities which may arise in relation to any of its electronic communication. No obligation is entered into by virtue of this e-mail, unless confirmed in writing and duly signed by an authorized representative of Globo. All outgoing and incoming emails are scanned for malicious code and all malicious code infections are being automatically reported to the appropriate IS Administrators.

On Sun, Sep 16, 2012 at 10:24:54AM +0300, ioakimoglou wrote:

Hello everyone, This is my first post in any of the RIPE mailing lists, I'm not even sure if I'm qualified or even "allowed". So feel free to put me in place, if needed :)

Anyone who uses internet resources (in the RIPE region) is the constituency of the apwg mailing list, so don't worry about being "qualified" :)

I find Randy's post about the last /8 going to new entrants to be the most convincing one. Giving it away to us (LIRS who are already in possession of at least a /21) should be secondary to giving it to newcomers who really need it in order to get started. I don't mean that's the only group of companies/orgs that should be considered, but they DO need this the most.

This is out-of-scope for this particular proposal though as it doesn't address whether ipv4 resources should be restricted to newcomers only. rgds, Sascha Luck

On 9/16/12 5:54 PM, Max Tulyev wrote:

Hello everyone,

I support this.

Also I can say the PI should be distributed as well. There are lot of companies, educational entities, small networks can't become a LIR as for money and/or formalities (bueracracy) issues.

...and all of them needs to be multihomed? Cheers, Jan

Hi,

Vitaly Turovets wrote Sun, Sep 16, 2012 7:14 PM 2012/9/16 Jan Zorz @ go6.si <jan@go6.si>:

...

On 9/16/12 5:54 PM, Max Tulyev wrote:

...

Hello everyone,

I support this.

Also I can say the PI should be distributed as well. There are lot of companies, educational entities, small networks can't become a LIR as for money and/or formalities (bueracracy) issues.

...and all of them needs to be multihomed?

Cheers, Jan

At least, some percent of those needs to.

Vitaliy Turovets

How about limiting IPv4 PI assignments to a total max of /12 (?) within the remaining /8 pool? Without knowing the exact statistics on IPv4 PI assignments by RIPE I guess this could be a compromise for entities needing IPv4 PI address space within the next time by still preserving enough resources for new members for the coming years. To higher the burden of possible abuse, IPv4 PI assignment requests may be only possible after having routed IPv6 PI address space, already. -Florian -- IT Manager, Dipl.-Ing. Florian Fuessl ----------------------------------------------------------------------- DegNet GmbH GF: Lothar Mayer Tel +49-991-3830566 Westl. Stadtgraben 20 AG Deggendorf HRB 2199 Fax +49-991-3830567 94469 Deggendorf http://www.degnet-gmbh.de mailto>>ff@deg.net<< -----------------------------------------------------------------------

On 17/09/2012 10:47, Gert Doering wrote:

... I'm not sure if "every time the available pool for X is empty, just grab another slice of the rest and declare it to be 'available for X, but limited!!'" is a reasonable strategy...

+1 To be clear, as proposer of this policy, I'm interested in a long term solution for PI so that we have long term policy clarity for PI assignment in the RIPE service region. I'm not interested in a series of stop-gap measures. Nick

On Mon, 2012-09-17 at 13:02 +0200, Richard Hartmann wrote:

On Mon, Sep 17, 2012 at 12:25 PM, Nick Hilliard <nick@inex.ie> wrote:

...

long term policy clarity for PI assignment in the RIPE service region.

Isn't it pretty clear now as it is? There will be no further PI assignments. PI is dead. Holders of PI have the option to: A) Keep holding, B) Return the address space (to become PA) C) Transfer it via a very complicated business procedure to another holder If they ever pick B, the address space returns to the non-available pool to be allocated again according to final-/8 procedures, to LIRs. (Please correct me if I'm wrong.) Ie., the long-term status of PIv4 is pretty bleak at the moment.

Strong agreement on both from me.

I support this PDP precisely because it enables fair run-out without arbitrary limits or temporary stop-gaps.

The run-out you refer to is of the final-/8 pool, some specific space. But, assignments and allocations aren't permanent by design, and right now we're lacking in the PI transfer area. Any space returned to RIPE cannot become new PI. One result of the current policy is that there will be no more PI assignments. Obviously it's one way to remove the distinction between PA and PI I guess, by simply removing PI as a concept. :) Two fixes comes to mind: 1) 2012-04, required if we want to keep the idea of PI - otherwise we should explicitly decide on removing the distinction... 2) A transfer-policy covering C) above, i.e fix another RIPE<->Reality disconnect I'd like to see both of them. Best, Martin

Hi, On Sun, Sep 16, 2012 at 06:54:49PM +0300, Max Tulyev wrote:

Hello everyone,

I support this.

Support *what*? The posting you've quoted argues for a completely different change ("do not give last-/22 to LIRs that already have space"), which is out of scope for the 2012-04 discussion. So your comment is somewhat unclear regarding support or non-support for the proposal at hand. 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 17.09.2012 05:27, Randy Bush wrote:

gedanken experiment

o cut last/8 allocations to lirs back to /24 Most likely too small to do anything meaningful with ...

o give end sites /29s from a specific block. lets them nat. Apart from uses like VPN transfer networks (for which a /29 should be

Anyway, I've been wondering for a while - how many new ISPs (or LIRs, for that matter) have been founded over the last - say - 2 years? Given the technical requirements as far as performance and bandwidths go, and the low end user prices for Internet connectivity, I don't see how there are any feasible business models to start an ISP business nowadays... this poses the question whether saving v4 addresses for new ISPs/LIRs is even relevant ... plenty sufficient in many cases, though possibly too smal in some special situations with many servers), this isn't sufficient for the predominant use, multi-homing. Question is - as others have already stated - whether multi-homing is a valid reason for getting v4 addresses in this phase ... I reckon if the policy should be altered to allow v4PI assignments, should there be a quota between PI and LIR assignments? I've not read up on the amount of reserved space for future uses, but say that's 20%, should there be a 4:1 ratio between PA and PI, eg. 64% PA and 16% PI? Also, should there be some sort of regulation as to what businesses ought to be allowed to receive LIR status? Otherwise, regular end customers with enough money (or enough despair) could chose to become an LIR, wasting more space than they would getting (currently unavailble) PI space, with only one company profiting from /22, while "real" LIRs might not get anything at some point, with dozens or hundreds of end customers not being able to get v4 addresses ... -garry

On Mon, 2012-09-17 at 08:40 +0200, Garry Glendown wrote:

On 17.09.2012 05:27, Randy Bush wrote:

...

gedanken experiment

o cut last/8 allocations to lirs back to /24 Most likely too small to do anything meaningful with ...

Anyway, I've been wondering for a while - how many new ISPs (or LIRs, for that matter) have been founded over the last - say - 2 years? Given the technical requirements as far as performance and bandwidths go, and the low end user prices for Internet connectivity, I don't see how there are any feasible business models to start an ISP business nowadays... this poses the question whether saving v4 addresses for new ISPs/LIRs is even relevant ...

Any future upstart-ISP *NEED* IPv4 space to do CGN. A small block is sufficient. Not having access to it is tricky w.r.t competition legislation (the holders of IPv4 not selling some to new-comers, thus blocking the market or whatever). This isn't an argument pro or con PIv4-from-last-/8 in itself, unless ISP operations are restricted to companies that are also LIRs, which, well, it isn't. For the record-keepers, I'm pro 2012-04. In my opinion, a real solution to the above problem is not found in the distribution model of an extremely limited amount of space (last /8). And we're going there anyway (to the situation requiring the real solution), I would just like us to get there sooner rather than later so we can stop fumbling for non-solutions. It gets us IPv6 faster, is my theory. And it's of course debatable whether new-LIR cost for /22 contra PI-process for /24. The opportunistic PI-applications are probably quite limited in volume? However, need for address space is somewhat in-elastic, like having gas in your car when driving to work. Many enterprises need it regardless of its form and will easily pay 3.5 EUR/address for one or several /22. FFWD, plz. Best, M

Hi, On Sat, Sep 15, 2012 at 04:57:48PM +0200, Sebastian Wiesinger wrote:

I agree with Randy and oppose this proposal. We should not use the last /8 for PI space but for LIRs who need at least a bit of v4 space to get going.

There is no distinct difference between LIR and non-LIR except paying two grand a year. Lets assume the simple scenario that someone wants IP Space. Without the proposal you pay 2 grand a year and get a /22, no questions asked. With the Proposal you pay whatever your LIR bills you and get a /24. I assume there will be quite some (non-ISP) Organizations who will choose being LIR over getting a PI because they get more IP Adresses. Without the proposal you force ALL organizations to do so. Now, which scenario will likely run out quicker? Rico MfG/regards, -- Rico Gloeckner teamix GmbH Suedwestpark 35 90449 Nuernberg Amtsgericht Nuernberg, HRB 18320 Geschaeftsfuehrer: Oliver Kuegow, Richard Mueller

On Sun, Sep 16, 2012 at 04:52:11PM +0200, Jan Zorz @ go6.si wrote:

IPv4 PI was never meant for that, but more for enterprises in order to be able to multihome. That luxury times are over and I think that the policy should reflect that.

that would be the LIR-centric view. However, I understand the intention of the LIR system for the LIRs to be resource-management organisations rather than a club with exclusive access to resources. Therefore an end-user is no less "entitled" to IP resources than a LIR and 2012-04 in current form is arbitrarily discriminationg. rgds, Sascha Luck

* Jan Zorz @ go6.si

I don't see much of a benefit for the Internet community and infrastructure if a company with 5 employees and two internet servers gets the whole /24 just because they need multihoming and are able to pay few EUR per year for that.

Currently they'll have to get a /22 instead. How is that any better? -- Tore Anderson Redpill Linpro AS - http://www.redpill-linpro.com/

On Mon, Sep 10, 2012 at 04:14:37PM +0100, Nick Hilliard wrote:

This is an extraordinarily discriminatory position.

In fact I find it not just to be unfair, but dramatically anti- competitive as the LIRs could be seen as harbouring the remaining IPv4 address space for members of their own club.

Good point, if someone were to take that to court or the competition commissar.

By rejecting this proposal, this discrimination against End User requirements will be permanently enshrined in RIPE policy and that they won't get the opportunity to apply for reclaimed address space in future. Again, I find this to compound the implicit unfairness of excluding them in the first place.

Possible compromise: make *returned/reclaimed* PI assignments available for assignment as PI rather than using them to make up patchwork PA allocations... rgds, Sascha Luck

On 9/10/12 10:34 CDT, Sascha Luck wrote:

On Mon, Sep 10, 2012 at 04:14:37PM +0100, Nick Hilliard wrote: ...

...

By rejecting this proposal, this discrimination against End User requirements will be permanently enshrined in RIPE policy and that they won't get the opportunity to apply for reclaimed address space in future. Again, I find this to compound the implicit unfairness of excluding them in the first place.

Possible compromise: make *returned/reclaimed* PI assignments available for assignment as PI rather than using them to make up patchwork PA allocations...

Note that besides addresses returned or reclaimed by RIPE there is the Recovered IPv4 Pool administered by the IANA, see the global policy [GPP-IPv4-2011, AKA RIPE-2011-01]; http://www.icann.org/en/news/in-focus/global-addressing/allocation-ipv4-post... And it has more than a /8 in it already; https://www.arin.net/announcements/2012/20120611.html So there are going to be dribs and drabs of IPv4 for a long time. (For those not familiar with the idiom "dribs and drabs", it means "small sporadic amounts.") There will not be enough to really make a fundamental difference, however there will definitely be more than enough for people to argue over, its human nature and is already happening. As I don't represent any resources in the RIPE region, I will not express an opinion on the policy itself. But I though it was import for people to realize there is more than a trivial amount of resources in the IANA Recovered IPv4 Pool even though it won't save the world from IPv4 exhaustion either. -- =============================================== David Farmer Email:farmer@umn.edu Office of Information Technology University of Minnesota 2218 University Ave SE Phone: 612-626-0815 Minneapolis, MN 55414-3029 Cell: 612-812-9952 ===============================================

Hi Nick,

It's also important to realise that the term "last /8" is now jargon for "all IPv4 addresses which the RIPE NCC handles after the exhaustion of the second last /8". In other words, it explicitly includes all IP addresses which are recovered by the RIPE NCC in future.

By rejecting this proposal, this discrimination against End User requirements will be permanently enshrined in RIPE policy and that they won't get the opportunity to apply for reclaimed address space in future. Again, I find this to compound the implicit unfairness of excluding them in the first place.

The change in the policy is for assignments from the last /8. If the policy would state that PI assignments are still possible from reclaimed address space (but not from the final /8 - the 185.x.x.x/8 range) I might reconsider. In the current form I do not agree with the proposal. Erik

Gert Doering wrote:

Hi,

On Mon, Sep 10, 2012 at 05:40:47PM +0200, Erik Bais wrote: [...]

...

In the current form I do not agree with the proposal.

Let me chime in here with the same statement, but emphasising the *in the current form* and let me add *in the current IPv4 policy framework/environment*

Noted, and we might want your input in wording this accordingly, if this is the compromise we can eventually agree on :-)

Here's _my_ reasoning: - in priciple I do see the arguments in favour and I tend to sympathise with them, because this would - potentially - remove an "unfair" imbalance, and there were substantial discussions already towards removing the distinction between PA and PI. but - as the proposal is worded right now, we would add more "special cases" and, as it was pointed out already, financial uncertainties or new and creative, business incentives. - if my information is correct, then addresses from the L/8 would *not* be restricted when it comes to transfers! This may just add inentives to grab addresses from the L/8 and sell^Wtansfer others from one's pool. [ Btw. this issue is already there, without 2012-04 passing. ]

The whole wording surrounding the "last /8" policy in the current documents has caused quite a bit of confusion in the past - but indeed it was agreed by the WG that as soon as this policy goes into effect, it's lasting, and "the last /8" will be "all of it, after the threshold has been crossed once"...

- exactly, and as I was already chastised for pointing out some of the omissions, imbalances and weirdnesses of the already existing policies (in a different discussion thread), while the substrate and the interpretation or validity of the "original intents" is not stable,

Gert Doering -- APWG chair

I do not see any other option for me than stating my *non-support* for this proposal. Going forward with this one, without amending some others at the same time, would just make the patchwork system approach worse and put us deeper into the ditch. I would be more than happy to change my position, as soon as there is a holistic attempt to improve the overall consitency of the different (existing) policies and the concurrently active proposals. Wilfried