Keyboard Shortcuts
Thread View
j: Next unread messagek: Previous unread messagej a: Jump to all threadsj l: Jump to MailingList overview
Dear All,
I hope my mail find you in good mood. First, I'd like to thank Mr.Toshiyuki Hosaka for presenting our proposal "Global policy for the allocation of the remaining IPv4 address space" in APNIC region.
Second, I'd like to reply to all comments on this proposal, I know, I'm late but excuse me I was under work pressure that prevent me from following up most of mails...
Philip Smith:
How are you doing Philip? I hope you be fine. I'm sure you have a great sympathy for this proposal (Thank you), But I've notice some contradictions in your emails, you wrote:
The problem for our region is that it brings the timeline of run out of address space for APNIC forward by a good 12 months, meaning just over2 years of IPv4 space left.
In any event, I believe at least RIPE NCC and APNIC are requesting smaller IPv4 /8 blocks from the IANA now, which goes a long way to ensuring the fairer distribution of what is becoming a limited resource.
First, you refer to the probable disadvantage of the proposal on APNIC which will forward the run-out time by 12 months, keeping in mind that APNIC in that time will receive 5 /8 blocks automatic from IANA, and APNIC could then manage assignment process in suitable manner as they want. Then you mention that APNIC & RIPE now will request smaller /8 blocks from IANA, which will lead to delay the run-out of IP blocks … (forward & delay) That’s one.
Regarding IPv6, All know about AfriNIC and LACNIC activities for promoting IPv6 in their regions. You specifically have participate in many training and workshops for IPv6 in AfriNIC region. And the best evidence on our activities is the last press conference held by AfriNIC and Adiel stated the AfriNIC activities toward that: AfriNIC has started since 2005 a campaign, which will be intensified, to raise awareness on IPv6 trough technical trainings and experience sharing events. Only 33 Operators have IPv6 addresses allocated in Africa region. We need by 1/1/11 to have IPv6 allocated in all countries in Africa.
If some of the LIRs in those two regions converted from NAT and double NAT to using real IPv4 addressing, both LACNIC and AfriNIC would receive more IPv4 /8 blocks sooner, making the chance of
them running out first less of a likelihood.
Again as Ra?l said, It is not LACNIC's objective. We don't want to promote a competition for getting IPv4 addresses from the unallocated pool. Right the opposite.
In addition to what I’ve said in last AfriNIC meeting in Abuja that if we do that it’ll become a battle on IP addresses and it’s not the aim .. we aim to equal fair distribution of the remaining IPv4 pool for all RIR … we had enough battles in many fields.. And I’d like to state it clearly that AfriNIC doesn’t discourage any LIR from applying for IPv4 space and use NAT instead !!!
NAT is not the core of the proposal so please we can discuss it latter.
Philip>As I said at the AfriNIC meeting a few months ago, it would be better if
LACNIC and AfriNIC made it very clear to their members (e.g. like ARIN Board did re IPv6)
Roque reply > Two things, first big address consumption ISPs at the bigger RIR
regions will not base their business plans on the possibility of getting addresses at LACNIC or AFRINIC, they will transit to IPv6.
Philip>The former is very possible. The latter still sees a lot of business
needing persuading to make the investment.
So let's not given them any reason to be creative - a uniform global "run-out" ensures that people can't be creative, and that they then can only consider IPv6 as part of their business model moving forwards.
Sorry, I got confused .. Do you encourage deploying IPv6 or not ? And from financial & investment for deploying IPv6, IMHO, ISPs in developed countries at bigger RIR regions are capable to support this investment more than others ISPs in developing countries..
Roque > You are missing that at the bigger RIR there are huge amount of legacy allocations and unallocated space that will be
fueling a secondary address market. Companies in less developed part of the world will have zero influence at this market.
Philip >I didn't see this discussed in the policy proposal. Besides, I don't see
how a legacy allocation market is going to help a newcomer. The Internet is growing very quickly in South and Central Asia, the Middle East etc. Or are they all meant to spend e-bay prices to get address space transferred?
The implicit mean of the proposal is to prevent monopoly for bigger RIR in selling their legacy allocated space which are not assigned for LIRs yet, by preserving a suitable amount of /8 blocks for each RIR to make use of it as per their own policies. And for fast growing Internet market such as in South & Central Asia, and Middle East if the allocated 5 /8 IPv4 blocks at their RIR’s run-out they can deploy IPv6. IMHO, It’ll be fair instead of inequity for the other slow growing market in other regions.
Roque > considering one of the biggest benefits of this policy. It release the
pressure on the central IANA pool allowing each RIR to focus on their own policies.
Philip>I don't see this at all. How would it reduce pressure? And how would it
allow the RIR (and presumably the RIR membership) focus on their own policies? What would be needed to be focused on?
It’ll reduce pressure by preventing battle on the remaining IPv4 blocks, and will allow RIR to focus on their policies and how to modify it if needed for distributing the allocated blocks from IANA.
I've even had people ask me if they can "sell" their IPv4 address space when it becomes valuable in the next year or two and use NAT instead.
That’s assuring the need for such proposal to avoid as much as we can monopoly and exploitation. So it’s better to reach an equity distribution of the remaining pool.
Ra?l > But I only wanted to point out that it is not
about LACNIC or Afrinic, so saying LACNIC and/or Afrinic shoud do "something" is a wrong approach to the discussion.
Philip>Agreed also. Everyone needs to do something, but I think we need to
really figure out what this "something" is. I don't think prop-51 is it, unfortunately.
So, what’s this suitable something in your opinion? Is it deploying of IPv6 .. It’s determinism, we all admit that, and this proposal doesn’t discourage that… It’s only seeking for the fair and equal allocation of the remaining IPv4 pool to all RIR and each RIR is free to assign its allocated blocks according to its existing or modified policies. For example may RIR reserve part of its blocks for critical operation assignment or modify the existing assignment policy to enforce the newcomer LIR to have IPv6 deployment before applying for IPv4..
Yes; there are currently many organisations who have RIR membership in more than one region. If they need more IPv4 address space, they will apply for it. Not doubting the RIR staff for a second, but we can't assume for a second that applicants won't be creative.
What is the percentage of these ISPs with regard to the whole ISPs in all RIR regions ?
So this proposal basically says to APNIC members "hey, give up one year's worth of /8s so that LACNIC & AfriNIC don't have to worry about IPv4 space for another 3 or 4 years hence". Where is the advantage for the APNIC membership? Why would APNIC members request APNIC to >change allocation policies, and what to?
Well, very good question.. Let’s consider the both situations, keeping the presence on demand IANA allocating policy and the proposed policy.. IANA now have about 46 /8 unallocated IPv4 blocks, In case of on-demand policy any RIR probable allocations will ranged from 46 blocks in best case –to- Zero blocks in the worst case (no allocations) that depends on LIR requests during the coming three years as statistical said IANA pool will exhaust in 2010 .. and In case of the proposed policy any RIR probable allocations will ranged from 26 blocks in best case –to- 5 blocks in worst case..…. What do you think ? Is loosing 20 blocks allocated for other RIRs in the second case is compared with zero blocks in the worst case in the first case? Keeping in mind that these 20 blocks you could loose in the first case as they could be allocated to other RIRs upon their request.. but in the first case you may have Zero blocks or less than the five blocks the proposal asking for.
David Conrad: Hi, It’s nice to meet you here in the mailing list and hope to see in the coming event ISA.
Indeed, but it isn't clear to me how this proposal helps that situation. The ISPs with the most power are also the ones who probably already have offices/subsidiaries/partners/etc. in Latin America and Africa. Unless AfriNIC and LACNIC become _extremely_ stringent on membership and invest heavily in verification mechanisms, I don't see the larger ISPs even blinking at this sort of thing. Just the cost of doing business...
Also the same question: What is the percentage of these large ISPs that have many offices in different RIR regions with regard to the whole ISPs in all RIR regions ? and to avoid the probable second market for IP addresses it’s better to insure at least 5 /8 blocks for each RIR. And each RIR is capable for regulate distribution and assignment of its pool in order to avoid that situation as much as possible… for Egypt there is three Large ISPs has offices in other RIR than AfriNIC and they do use their IPs and AS they got from AfriNIC.
Roque >Two things, first big address consumption ISPs at the bigger RIR regions
will not base their business plans on the possibility of getting
addresses at LACNIC or AFRINIC, they will transit to IPv6.
David>No they won't. They will do whatever is necessary to obtain the IPv4 >address space they need to continue business.
ISPs in developed countries will do whatever is necessary to obtain the IPv4 to continue their business and not transit to IPv6, whereas ISPs in developing countries which are less in technical capabilities and face financial problems are required to transit to IPv6 faster than ISPs in developed countries .. Is it the fairness from your point of view ?
Rearranging where the IPv4 free pool sits isn't going to help things all that much >(although it might remove IANA as the target for lawyers, thanks! :-)). Large >scale ISPs have the resources to set up offices in Latin America and Africa (and >the resources, far more than those NICs do, to bury the NICs under paperwork >to justify their requests). The folks who will lose are the smaller ISPs in the >regions served by the larger NICs who lack those resources. The winners will >be those with lots of IPv4 addresses.
If it’ll not help, at least will keep fair & equity allocation of the remaining pool over existing RIRs and then each RIR could manage its allocated pool according to their policies/membership state. And as Roque said you can discuss the number of allocated /8 blocks for each RIR..
David>a) customers don't want IPv6 (nor do they want IPv4 -- they want "the >Web"/their pr0n, they don't care about the details).
b) migrating to IPv6 has real costs and because of (a), there are no additional >revenues to cover that cost.
c) they _can't_ migrate because their equipment/software vendors don't yet >support IPv6.
d) etc.
Roque > I agree, even if ISP do migrate to IPv6 there is not content there. So, we will have a double stack scenario for several years/decades.
I agree too and like to add that not yet all software applications & hardware appliances support IPv6 such as many firewalls and IPS. So IPv6 fully deployment will need a time may be long specially in developing countries.
Unless AfriNIC and LACNIC become _extremely_ stringent on membership and invest heavily in verification mechanisms,
Ra?l >Why not ? good point. But not only that. Other measures will be
necessaries in the future too for avoiding or limiting the RIR shopping. And this is something that should be done by all the RIRs due to the fact that nobody know which one will be the first in running out of IPv4 addresses (if one). And it not depends only in the distribution of the unallocated pool, but also on the regional policies for dealing with the last part of the regional stocks in each region.
I agree and as mention before each RIR can modify its policies in a manner to face this situation, with stressing on deploying IPv6.
P.S. It might also be argued that the paradox you note could be a blessing in disguise as it means those in developing countries will make the shift to IPv6 that much sooner.
Ra?l
This is something similar to say that developing countries will be the first ones in >shifting to ethanol based cars because they will not have oil. But so, we will >have to produce our own cars, and it will be of course more expensive than >buying the cars produced for the big mass. Shifting to IPv6 is an objective and >of course, it is probably more important for developing countries since they have >less margin for speculating, and of course we are working very much on that >and spending large amount of resources as we have been doing for the last 4-5 >years, but the main objective for all of us should be, IMHO, a non traumatic >transition to IPv6 worldwide.
Totally agree .. Transition to IPv6 will not happen an a day and night, It needs time not a short time but we have to start in same time we have to ,IMHO, be fair in allocating the remaining IP blocks in equal manner between RIRs. I’ve an example for IPv6 deployment in Egypt, we have 4 IPv6 /32 blocks (3 for ISPs and one for Ministry of Comm.) since 2004 and till now for ISPs no customers request for IPv6 (except the Universities/Research centers MPLS network), as some claim that not all their software support IPv6 and also hardware and others waiting till the Internet community shift to IPv6 then they will shift as they don’t want to live in isolated islands.
If RIRs were to encourage the use of NAT in such cases, the demand on the >remaining IPv4 free pool would be lessened, thereby extending the runway for >IPv6 deployment and (assuming NAT is the evil many say it is), encouraging >that deployment.
Please No NAT encourage, Philip can reply better than me for NAT advantages and disadvantages.. but Also NAT cost a lot of money as it requires special hardware cards (expensive one) in some routers to be able to do NATing
Randy Bush, How are you doing ? I hope you be in a very good health. I do benefit from IXP discussion in last AfriNIC meeting and hope you could attend the next MENOG meeting, It’ll be a session for peering & IXP in the middle east.
and we, the users, want to limit rir shopping, why? it may be the only bit if >market competition there is in the rir monopoly game. some may see one good thing about the iana ipv4 free pool run out as the creation of an actually visibly competitive market in address space.
I think, IMHO, we have to avoid RIR shopping to avoid RIR monopoly which will allow for exploitation of the situation, by applying this proposal to equal & fair allocation of the remaining pool in conjunction with promoting IPv6 deployment.
Geoff Huston
Most of the "transition" plans we have seen so far for this transition to an IPv6 >network pass through this intermediate stage of dual stack deployment. The >basic idea is that what is required for a V4 host initiate, maintain and close a >"conversation" with a V6 host and vice-versa goes well beyond the conventional >mode of packet protocol header substitution, and efforts to perform various >permutations of protocol header translators, DNS manipulations and application >level gateways all appear to have their dark and ugly side in terms of cost, >deployment complexity, service fragility and fractured application transparency.
So, with help from various forms of tunneling support to bridge over any >protocol-specific transport continuity gaps, the basic idea of this transition is that >we enter an extended period where hosts need to use V4 to talk to V4 hosts >and V6 to talk to V6 hosts. As long as hosts first try the V6 handshake, then, so >goes the line of reasoning, we should see the traffic mix in this dual stack tend >to move to v6 as the legacy V4 infrastructure migrates to V6, and the dual stack >nature of the deployment means that dual stack hosts can fall back to V4 to >speak to V4 legacy infrastructure.
I do agree with you in that. Thank you Geoff for this declaration, that we all need IPv4 and this proposal is a trial to make the last IANA IPv4 allocations fair and equal for all RIRs whether for developed region or developing regions....
At the end of this long email, I’d like to emphasis that this proposal is not against the deployment of IPv6 at all.
Regards, Haitham EL Nakhal
P.S. ISA = God willing.
Hello Hytham,
(I've snipped out the bits and pieces not directed at me - you are right, it was a very long reply of yours! :-))
Hytham EL Nakhal said the following on 12/8/07 02:16:
How are you doing Philip? I hope you be fine. I'm sure you have a great sympathy for this proposal (Thank you), But I've notice some contradictions in your emails
First, you refer to the probable disadvantage of the proposal on APNIC which will forward the run-out time by 12 months, keeping in mind that APNIC in that time will receive 5 /8 blocks automatic from IANA,
I wasn't aware that IANA automatically gave address blocks to the RIRs. Last I heard was that the RIRs had to make a pretty strong application case for getting another block.
and APNIC could then manage assignment process in suitable manner as they want. Then you mention that APNIC & RIPE now will request smaller /8 blocks from IANA, which will lead to delay the run-out of IP blocks … (forward & delay) That’s one.
Umm, I never said that requesting smaller blocks will delay the run out. A tank with so much water in it has so much water, whether the tap is wide open or only trickling.
prop-051 will unquestionably bring the run-out forward for APNIC's service region.
Regarding IPv6, All know about AfriNIC and LACNIC activities for promoting IPv6 in their regions. You specifically have participate in many training and workshops for IPv6 in AfriNIC region. And the best evidence on our activities is the last press conference held by AfriNIC and Adiel stated the AfriNIC activities toward that: AfriNIC has started since 2005 a campaign, which will be intensified, to raise awareness on IPv6 trough technical trainings and experience sharing events. Only 33 Operators have IPv6 addresses allocated in Africa region. We need by 1/1/11 to have IPv6 allocated in all countries in Africa.
Allocations are only the first small step. How many of these /32s do we actually see in the routing table today? That's the problem which people need to address.
Again as Raúl said, It is not LACNIC's objective. We don't want to promote a competition for getting IPv4 addresses from the unallocated pool. Right the opposite.
prop-051 proposes competition. When one shop doesn't have bread, you go to another shop to get bread. prop-51 says that AfriNIC and LACNIC can have bread for several years to come, whereas RIPE NCC, APNIC and ARIN have to have cake instead. People who don't want cake will do anything they can to get bread. Whether they go to LACNIC & AfriNIC, or go to ebay, etc... This is a given, unfortunately.
In addition to what I’ve said in last AfriNIC meeting in Abuja that if we do that it’ll become a battle on IP addresses and it’s not the aim ..
There will be a battle, sadly. There is very little the RIRs or anyone else can do about it.
we aim to equal fair distribution of the remaining IPv4 pool for all RIR … we had enough battles in many fields.. And I’d like to state it clearly that AfriNIC doesn’t discourage any LIR from applying for IPv4 space and use NAT instead !!!
I hope no RIR does - but then again, I know from the wider industry that many ISPs believe that the RIRs encourage the use of NAT. But that's neither here nor there for this proposal.
Sorry, I got confused .. Do you encourage deploying IPv6 or not ? And from financial & investment for deploying IPv6, IMHO,
I encourage the deployment of IPv6 on existing networks, have been doing this for many years. We're going to have to live dual stack for several years to come, like it or not. Those who do not have real IPv4 addresses by the time the IPv4 pool runs out will either have to NAT or start paying big money for real IPv4 addresses.
The implicit mean of the proposal is to prevent monopoly for bigger RIR in selling their legacy allocated space which are not assigned for LIRs yet, by preserving a suitable amount of /8 blocks for each RIR to make use of it as per their own policies.
I didn't see any proposal about the RIRs selling legacy address space? Can you point to where this has been proposed, or announced?
In the APNIC region, it is a very bad idea to imply anything in any proposal. It needs to be written out specifically - anything which is not spelled out does not happen.
And for fast growing Internet market such as in South & Central Asia, and Middle East if the allocated 5 /8 IPv4 blocks at their RIR’s run-out they can deploy IPv6. IMHO, It’ll be fair instead of inequity for the other slow growing market in other regions.
Sorry, I don't understand why you think it is fair for the world economy to put something to the side for the slow providers to finally get around to getting their act together. If people want to run an ISP, they should get on and do it. Up to 2010 or so, they'll get IPv4 address space (and need a really good plan of action to get IPv6 deployed as well, for their own sakes). And after that they'll get IPv6 address space from the RIRs. Same as everyone else. If they need to access the remnants of the IPv4 Internet, that's where IPv4 NAT will have to serve the need. Unless someone manages to undeprecate NAT-PT.
It’ll reduce pressure by preventing battle on the remaining IPv4 blocks, and will allow RIR to focus on their policies and how to modify it if needed for distributing the allocated blocks from IANA.
There is going to be a battle regardless. That's why I think that prop-51 is simply tinkering around the edges and is doing little to solve the real issue. ;-)
That’s assuring the need for such proposal to avoid as much as we can monopoly and exploitation.
So it’s better to reach an equity distribution of the remaining pool.
The current distribution system is as equitable as it gets. I don't see how anyone can claim that come 2010, all 5 RIR regions will have exactly the same numbers of LIRs and exactly the same address space requirements.
So, what’s this suitable something in your opinion? Is it deploying of IPv6
Leave the status quo with IPv4 distribution. And encourage every single LIR to put together a sane and sober plan to deploy IPv6 in their infrastructure.
It’s only seeking for the fair and equal allocation of the remaining IPv4 pool to all RIR and each RIR is free to assign its allocated blocks according to its existing or modified policies.
I think we have agreed that my definition of fair (every LIR gets an equal chance) and the proposal's definition of fair (some LIRs get more chances than others) are different.
For example may RIR reserve part of its blocks for critical operation assignment
What's critical operation? Who will define that? The LIRs who have the most money or influence within the RIR are the most critical? (Let's not start "what is critical" debate here.)
or modify the existing assignment policy to enforce the newcomer LIR to have IPv6 deployment before applying for IPv4..
Well, given the LIRs determine the RIR policy, I can't see LIRs voting for something like this, not currently anyway.
What is the percentage of these ISPs with regard to the whole ISPs in all RIR regions ?
Very small, but will increase markedly if prop-51 is approved globally...
Let’s consider the both situations, keeping the presence on demand IANA allocating policy and the proposed policy..
IANA now have about 46 /8 unallocated IPv4 blocks,
And there is a whole range of addresses outside the distinct /8 blocks in the former B space which are still vacant. About 1500 /16s the last I counted.
In case of on-demand policy any RIR probable allocations will ranged from 46 blocks in best case –to- Zero blocks in the worst case (no allocations) that depends on LIR requests during the coming three years as statistical said IANA pool will exhaust in 2010 ..
Yup, those are the predictions.
and In case of the proposed policy any RIR probable allocations will ranged from 26 blocks in best case –to- 5 blocks in worst case..…. What do you think ?
If 5 blocks are kept for each RIR, that means 21 would be left in the free pool. Oh, you mean that each RIR can get a maximum of 26 blocks under prop-51, with N set at 5.
Is loosing 20 blocks allocated for other RIRs in the second case is compared with zero blocks in the worst case in the first case?
Keeping in mind that these 20 blocks you could loose in the first case as they could be allocated to other RIRs upon their request..
but in the first case you may have Zero blocks or less than the five blocks the proposal asking for.
Say APNIC, RIPE NCC and ARIN consume round about 5 /8s per year each. If we had the 25 held for the RIRs idea, that'd leave 21 blocks. 15 /8s would give us one year, with just 6 blocks remaining after that. Divide those equally between those three RIRs. So we'd probably be 15 months from now with just the reserved /8 blocks left - which would give APNIC 15+12 months - 27 months from now to nothing left.
If we follow the current system, APNIC has 36+12 months to nothing left.
AfriNIC consumes one /8 per two years. Reserving 5 /8s for AfriNIC means that the African Internet can languish in IPv4 land for the next decade. While the rest of us have moved on.
Then replace AfriNIC in the above with LACNIC. Same is true.
And then we hear that AfriNIC and LACNIC are at the forefront of encouraging their LIRs to deploy IPv6.
Doesn't anyone see that the proposal completely contradicts that messaging??
I guess that's why I at least simply don't understand why the proposal is seen as something great and wonderful. I think it is the biggest tragedy that could befall the Internet in Africa and Latin America if it was globally approved.
So, if we are all being sensible, we should let it drop and move on doing real things like encouraging providers to think and plan how they are going to remain at the forefront of the networking business after 2010.
philip --
Hi,
Sorry for the slow response.
On Aug 11, 2007, at 9:16 AM, Hytham EL Nakhal wrote:
David Conrad:
Indeed, but it isn't clear to me how this proposal helps that situation. The ISPs with the most power are also the ones who probably already have offices/subsidiaries/partners/etc. in Latin America and Africa. Unless AfriNIC and LACNIC become _extremely_ stringent on membership and invest heavily in verification mechanisms, I don't see the larger ISPs even blinking at this sort of thing. Just the cost of doing business...
Also the same question: What is the percentage of these large ISPs that have many offices in different RIR regions with regard to the whole ISPs in all RIR regions ?
Probably fairly small. However, it is probably worth noting that the majority of address space at the RIRs is consumed by a small minority of the members.
and to avoid the probable second market for IP addresses it’s better to insure at least 5 /8 blocks for each RIR.
Why do you think 5 /8s for each RIR will avoid a secondary market?
And each RIR is capable for regulate distribution and assignment of its pool in order to avoid that situation as much as possible… for Egypt there is three Large ISPs has offices in other RIR than AfriNICand they do use their IPs and AS they got from AfriNIC.
I am not fully aware of the situation in Africa (nor in Latin America and the Caribbean), but I would be surprised to find that there aren't countries in which it is possible to establish shell companies that do little more than establish a legal presence in that country. Further, it doesn't even have to be a shell, it could as easily be a partnership with an existing company, including an established ISP. All it takes is a bit of money, which large scale ISPs generally have.
ISPs in developed countries will do whatever is necessary to obtain the IPv4 to continue their business and not transit to IPv6, whereas ISPs in developing countries which are less in technical capabilities and face financial problems are required to transit to IPv6 faster than ISPs in developed countries .. Is it the fairness from your point of view ?
No, I wouldn't consider it fair. However, from the perspective of the large scale ISPs, it is just business. In any event, the point I've been trying to make is that regardless of what the RIRs do, a secondary market will almost certainly establish itself and it would probably be best for the RIRs to figure out how to deal with that fact rather than pretending business realities don't apply.
Please No NAT encourage, Philip can reply better than me for NAT advantages and disadvantages.. but Also NAT cost a lot of money as it requires special hardware cards (expensive one) in some routers to be able to do NATing
Given IPv6 is not backwards compatible with IPv4, you'll need some form of network address translation to allow folks who are IPv4-only to talk to sites that are IPv6-only and vice versa. The future is _full_ of some form of NAT.
Regards, -drc
Activity Summary
- 5883 days inactive
- 5883 days old
- sig-policy@lists.apnic.net
- 3 participants
- 2 comments