Dear Mr. Agaral,
As promised, my reactions to the first part of your reply are inline below.
On Aug 27, 2011, at 3:01 PM, Tom Vest wrote:
Dear Mr. Agarwal,
While I remain strongly opposed to this proposal for many reasons, including those that you have specifically sought to challenge in this message, I thank you nonetheless for your very cogent and direct reply. IMO, the best and only legitimate way that discussions like this can contribute to the practical goal of policy decision-making by consensus is by directly engaging the various arguments for and against specific policy proposals -- their underlying logic, internal consistency, implicit assumptions, etc.
I will respond in kind in a separate message to follow shortly.
Even if the two of us remain at odds about this specific proposal forever, perhaps this exchange may still do some good by contributing to a return to that more constructive model of policy deliberation.
Sincerely,
Tom Vest
On Aug 27, 2011, at 1:53 PM, RAKESH MOHAN AGARWAL wrote:
Dear all,
Since my last posting to SIG Policy on 22/08/2011 I have received number of comments from participants. Through this reply I am giving my response to these comments. I would like to reiterate that the theme of prop-100 is to ask the APNIC community to make a policy for reservation of contiguous IP address blocks of certain minimum size for all countries in the APNIC region whether they are demanding or not so that they can meet out the requirements of their organizations out of these reserved blocks in future.
IPv4 distribution in 1996 versus today if proposal was accepted in 1996 (comments by Mark Newton, John Mann, Tom Vest etc.)
(a) If a policy like prop-100 would have been implemented in 1996, the IPv4 distribution today would have been very different. The status of IPv6 today in terms of less demand ntributing to and large availability is similar to the situation of IPv4 in the early nineties.
With respect to the *most relevant* way that these two situations might relate to each other, this is simply not true -- or alternately, it is only true if you are aware of some broad scientific consensus about the future trajectory of IPv6 demand that I have never heard of, and that is radically different from the views that I am familiar with.
For the sake of simplification, lets stipulate that by "the early 1990s," 50% of all available IPv4 had already been delegated. At that point in time, there was already widespread scientific consensus that the IPv4 addressing format could not accommodate a continuation of the empirically measured growth/demand trend for IPv4 addresses even for the foreseeable future. That widely shared, evidence-based consensus inspired a wide variety of responses, including private addressing set-asides, address translation, CIDR, interdomain routing, the adoption of "need-based" eligibility criteria and practices for distributing number resources, the RIR system of proxy-administration for some of those distribution processes, and ultimately IPv6.
In order for the "early 1990s" to be similar to the present, and for that similarity to support your argument in favor of Prop-100, there would need to be widespread consensus today that IPv6 is absolutely certain to be substantially or completely depleted in the foreseeable, i.e., not "possibly depleted in some indefinite/distant future." Do you subscribe to such a theory -- either your own or someone else's -- and if so, could you please point list members to some source for more details about it? Could you also provide some indication of the responses (ala "peer reviews") and level of support that it has received from members of the (presumably scientific) community in which this theory was developed?
Due to inadequate planning of IPv4, large blocks of IPv4 were indiscriminately given to different organizations in only a few countries, thereby creating shortage of addresses very soon.
When this fact dawned upon the custodians of IPv4 addresses, needs based policies were evolved to restrict the allocation of addresses.
With all due respect, this kind of "they should have known better then, because it harms us now" claim is not very useful. It's not even an argument so much as it is an extremely condensed and biased summary of a very long and complicated chain of historical outcomes. The problem is not that such claims are always without merit, but rather that they are always true to some degree, and not just in those cases when we happen to find it convenient to bring them up.
To illustrate, I note that India possesses the world's largest reserves of the element Thorium, which is (I believe) certain to become indispensable to various important global economic functions in the future. In that future, the majority of Thorium seekers around the world are likely to dismiss the fact that India was disproportionately blessed by nature with Thorium deposits as nothing more than an accident of history. Future generations, especially outside of India, are sure to wonder why "adequate planning" for future global Thorium requirements were not undertaken in the early 21st century, and are sure to demand remedies for that past (i.e., our present) injustice.
Now that you are aware of this future possibility, we may assume that it is within your power today to avert that outcome... but should we? What could you/should you do, even if you actually shared this vision of the future and genuinely wished to prevent future Thorium-related disputes?
The prices of IPv4 addresses went up and became costly for organizations which came later.
Let's put this assertion in context. With respect to all of the "organizations that came later" that actually possessed the means to put IPv4 addresses to the use(s) for which they were designed, the cost of IPv4 rose uniformly for all, and it rose to a degree that, in objective terms, can only be described as a tiny, tiny share of the cost of actually obtaining the means to put IPv4 addresses to good use. That was and will remain true for as long as the RIRs remain in the IPv4 allocation business. After that, this will still remain true for IPv6 and ASNs -- at least for address seekers that are permitted to obtain resources from an RIR, for as long as those very large resource pools endure.
The net result is that today late entrant organizations are paying a high cost for IPv4 addresses and in reality subsidizing those organizations who are holding more addresses.
This statement will not, and could not be true in India, for example, if incumbent Indian ISPs make their existing networks transparently accessible to late entrant organizations that possess only IPv6.
In fact, this statement can *only* be true if current Indian ISPs have failed to "plan inadequately" -- even for India's own future networking needs -- and persist in failing to plan adequately in the years to come.
(b) Regarding the comment that such a policy in 1996 would lead to more wastage of addresses upto 50% is hypothetical. Is there any study which leads to such conclusions?
I will look for related research. In the mean time RFC 1715 provides some useful empirical observations: http://tools.ietf.org/html/rfc1715
Those observations strongly suggest that for any address pool, the maximum level of "allocation efficiency" that is practically achievable is strongly determined by the level(s) of hierarchy that must be accommodated. Based on this insight, and all else remaining equal, it would seem that the introduction of yet another, national level of addressing hierarchy could only have resulted in even more "waste" and even faster depletion of the overall IPv4 address pool.
Even if we assume that there would have been more wastage, the beauty of a prop-100 like policy is that it would have led to the following –
a. All the countries would have got some address space reserved for their organizations for future use
http://www.un.org/en/members/growth.shtml#1990
In 1990 there were 159 recognized sovereign nations. Your proposal would have provided each with appx. one /8 + /9
http://www.un.org/en/members/growth.shtml#2000
The recognition earlier this year of South Sudan brings the current number of countries up to 193.
If IPv4 is truly as important as you suggest, your proposal would presumably have effectively doomed 44+ new sovereign entities to remain permanently excluded from the Internet.
b. Technologies like NAT was primarily developed to compensate the IPv4 address shortage which distorted the original architecture of the Internet
Actually the opposite would have been more likely. Unless you postulate that the mere possession of IPv4 addresses itself causes the process of Internet development -- including the spontaneous deployment of ubiquitous network facilities and the immediate possession of addressable hardware by a large share of all individuals and institutions regardless of income -- then your proposed distribution scheme would have done nothing more than cause the premature exhaustion of address resources in those places that were fortunate enough to have such endowments in the early 1990s. Given the quantities of addressing that your proposal would have provided, public IP addresses would probably have never been seen outside of the academic/R&E community. Under those circumstances, it seems unlikely that the Internet would have ever taken off, anywhere, and that the wealth of content, services, and opportunities that the Internet now represents would have never existed.
c. IPv6 would also have come much earlier.
Based on the chain of reasoning described above, I think it more likely that the conditions that caused the Internet to spread beyond the confines of relatively restricted R&E networks would have never developed anywhere. Rather than coming earlier, IPv6 would have never come at all.
d. There would be less of subsidization of organization in countries holding large blocks of addresses.
If large incumbent Indian ISPs elect to make their IPv4-based networks transparently accessible to other IPv6-only networks, who exactly would be subsidized by that action?
If large incumbent Indian ISPs elect to make their IPv4-based networks transparently accessible to other IPv6-only networks, in what way would future Indian ISPs that have only IPv6 be compelled to subsidize unnamed foreign IPv4-holdng networks?
If incumbent Indian ISPs decline to make their IPv4-based networks transparently accessible to other IPv6-only Indian ISPs, is somebody still being subsidized -- and if so, who is paying and who is collecting?
(c) Regarding the comments that hoarding of addresses by economies would lead to dampening of the internet / smart phone revolution is again hypothetical and perhaps unsustainable. E.g. a smart phone revolution is happening in India in spite of shortage of addresses. So there appears no correlation between the two.
I believe that you may have misinterpreted my previous point about how addressing can and cannot influence the pace of Internet development.
Hopefully my comments above about RFC1715, the number of countries in the world, and the more likely hypothetical results of early addressing scarcity will help to clarify.
If the scarcity of useful addresses reduces the likelihood that an industry will develop around opportunities to create rich/interactive content, then the imposition of such scarcity to all places would have reduced the odds that such an industry would have emerged anywhere at any point in time.
But perhaps we are using the same term to describe two entirely different things. When I think of "smart phones" in this particular context, I assume that we are talking about handheld devices that are intended to be capable of directly and transparently interoperating with other devices and other services that are administered by unrelated, potentially competing network providers. I believe that the smart phone revolution that is currently underway is driven in part by proven demand for the rich universe of content and services that emerged in a world where conditions of scarcity were *not* artificially imposed on IP addressing before the earlier "online content revolution" began. If early scarcity had pre-empted the development of the Internet outside of the R&E community, then that original explosion of content would not have happened, and the companies and end users that are now positively eager to invest large sums in smartphone deployment would never have had any basis for that enthusiasm.
I will save my reactions to the second half of your message for another message to follow shortly.
Regards,
Tom Vest
When 2 years ago, there were 300 million addresses, why Indian ISPs did not apply ? What is broken in the current APNIC policy ? (Comment by John Mann)
(a) First I would like to state that our proposal is for all countries in the APNIC region and not just for India alone.
(b) As far as allocation of IPv4 addresses to Indian ISPs is concerned, the commenter has tried to guess some reasons but those are just assumptions as no such study has been done by anyone on this subject as we understand.
(c) Two years ago India was having just 18.5 million addresses and now it has 34.5 million addresses for more than 350 million data users out of more than 850 million mobile users. There is almost 90% growth in allocation. This may be happening with other economies also. But it seems current APNIC policies are not taking adequate care to forecast needs of different economies to make provision for adequate address blocks for them. Therefore, I would like to summarize below what I feel is broken in current APNIC policy
If countries / organizations are not aware of their requirements then it means there is lack of awareness and APNIC needs to do something about it. Perhaps all stakeholders are not adequately involved. Only those organizations are being entertained who are approaching APNIC, others are possibly not been taken care of.
Late entrant organizations are paying a higher price for their addresses as compared to those who came earlier. By giving a suggestion “to go for reservation after 25% exhaustion of the IPv6 address space”, a situation like IPv4 is being encouraged in which organizations that will come at a later date will be made to pay a higher price than what it would cost them today.
(d) APNIC may like to set up some kind of a study to work out the details covering all the countries in the APNIC region and suitably address the issues in their policy.
Conflict between “Operational Policies of NIR” and prop-100 (Comment Chu-Yi)
It is possible that that there may be conflict between prop-100 and NIR policy. Existing policies may need review and changes made wherever needed.
Comments by M/s Reliance (Indian Service Provider)
They have suggested working out a mode of distribution of IPv6 addresses to different organizations if a large block is given. I would like to state that this mode of distribution can be left to the individual countries provided they are not in conflict with APNIC policies. As far as India is concerned it can be done by proposed NIR involving all stakeholders.
In addition to above, I will also like to tell that many technical comments have also been given to prop-100 and people have tried to examine this proposal entirely on its technicalities. But the purpose of proposal prop-100 is to ask the APNIC community to make a policy for reservation of contiguous IP address blocks of certain minimum size for all countries in the APNIC region whether they are demanding or not so that they can meet out the requirements of their organizations out of these reserved blocks. There is nothing much technical about it. Though I have mentioned some technical advantages, but these are not the sole purpose of the proposal. The proposal is all about judicious planning and distribution of addresses with a long term perspective.
(R.M.Agarwal)
(B.K.Nath)
sig-policy mailing list
sig-policy@lists.apnic.net
http://mailman.apnic.net/mailman/listinfo/sig-policy