Search RPD Archives
Limit search to: Subject & Body Subject Author
Sort by:

[rpd] IPv4 Soft Landing BIS

Owen DeLong owen at
Sat Jul 29 16:05:12 UTC 2017

> On Jul 28, 2017, at 17:13, Andre van Zyl <vanzyla at> wrote:
> Hi Owen,
> It might just be me, but I'm struggling to understand your actual position at the moment. 
> In the first part of your response below, you seem to argue for doing away with IPv4 as soon as possible, yet later, in your egg analogy, you seem to be making the case that restricting existing operators' access to further IPv4 resources is damaging their business?

Yes. Both are presently true. The sooner the internet moves away from this IPv4 dependency, the better it will be. The IPv4 dependency does, however unfortunately, exist today. So yes, you have summarized the facts surrounding my position accurately. 

> For the record, I am very much in favor of the move to IPv6 as far and as fast as is practical. However, there is no denying that IPv4 will not go away soon, so I am in favor of protecting IPv4 resources for new operators to be able to connect to the legacy internet, from an IPv6 network. Does that merit reserving the size of block as per this proposal? Perhaps not, but I think some needs to be reserved.

This is my point. This proposal doesn't protect resources, it prevents them from being used.

> See further comments in-line below.
> On Fri, 28 Jul 2017 12:02:53 -0700
> Owen DeLong <owen at> wrote:
>>> On Jul 28, 2017, at 04:18 , Jackson Muthili <jacksonmuthi at> wrote:
>>> On Fri, Jul 28, 2017 at 9:32 AM, Owen DeLong <owen at> wrote:
>>>>> On Jul 27, 2017, at 22:45 , Jackson Muthili <jacksonmuthi at> wrote:
>>>>> On Thu, Jul 27, 2017 at 1:16 AM, Owen DeLong <owen at> wrote:
>>>>>> Frankly, IMHO, the preservation of IPv4 is strictly a way of inflicting
>>>>>> additional cost and pain on the majority of the internet. Unfortunately,
>>>>>> much like toxic polluters of the 70s and 80s, most of the cost and pain is
>>>>>> inflicted on those who are ready for IPv6 rather than those who remain
>>>>>> unprepared for that future. The good news is that if the current adoption
>>>>>> rates continue, the holdouts that haven’t implemented IPv6 will become
>>>>>> mostly irrelevant relatively soon and when the rest of us start turning off
>>>>>> IPv4, they will be the ones left out in the cold wondering what happened
>>>>>> instead of inflicting costs and pain on the rest of us.
>>>>>> The sooner the internet moves on from its unhealthy IPv4 addiction the
>>>>>> better. I’m pretty sure you know this as well as I do, despite all of your
>>>>>> apparent protestations to the contrary.
>>>>> - Our region is still a young and growing region relative to yours.
>>>> Like it or not, Jackson, I’m part of this region, too. Nonetheless,
>>>>> - IPv6 is the end. But IPv4 is still a means to the end.
>>>> No… IPv4 is a temporary means of survival and is to some extent the status quo. Nothing more.
>>> It can be looked at as a temporary means of survival and that very
>>> premise makes it the means to the IPv6 end. Because the internet still
>>> operates mostly on IPv4 and an IPv6 only island cannot be reached to
>>> by an IPv4 only island unless there a mechanism to make the two
>>> co-exist has been applied. So semantics aside we seem to agree in
>>> principle.
>> No, the IPv6 end can be achieved without any IPv4 at all if one desires to do so.
> How would this IPv6-only network connect to the legacy IPv4 internet?

It doesn't. If we are at the IPv6 "end" being described, there isn't really an IPv4 internet to connect to. My point isn't that about connecting v6 and v4. It's about the fallacy that v4 resources are necessary or even particularly useful for deploying IPv6. There was a time when this was somewhat the case, but today, v4 resources are really only useful for communicating with the legacy internet and nothing more. 

>> IPv4 doesn’t do anything to help you deploy IPv6.
>> Therefore, IPv4 is not an means to the IPv6 end. IPv4 is a means to communicate
>> with the legacy internet. Nothing more, nothing less.
> Ah, so you kinda DO need IPv4 then.

I never said you didn't need IPv4 today. I'm saying it is becoming less and less necessary and making sure that there's a bunch of it still on the shelf when it becomes useless by denying it to operators with a clear and present need is simply bad resource management. 

>>>>> - If IPv6 were the absolute solution we would not have a booming a
>>>>> billion dollar IPv4 market.
>>>> That’s the same kind of logic currently being used by the climate change deniers.
>>>> In the 1960s and 1970s, in my other region, the argument was that if dumping toxic chemicals
>>>> into waterways was a real problem, it wouldn’t be so cheap to do so. Fortunately, the
>>>> EPA was created and huge fines were put in place and the superfund was created to try and
>>>> shift some of the costs of these toxic waste dumpings back on to the sources instead of
>>>> the down-stream victims.
>>> I don't think we should parade IPv4 at the same level as toxic and
>>> hazardous waste.
>> I’m not. I’m parading the use of IPv4

That should have bee placing instead of parading. Damn autocorrect. 

>> combined with the failure to implement IPv6
>> at a similar level because it has the same kind of cost-shifting effect.
>>> But using your analogy, please help me understand how the two liken,
>>> who is dumping what on who, and which side is facing any costs as a
>>> result.
>> By failing to implement IPv6 and continuing to operate IPv4 only, an organization is
>> forcing everyone else that wants to communicate with them to continue to maintain and
>> in some cases expand their own IPv4 infrastructure at an ever increasing cost, much the
>> way that those dumping toxic waste were saving money by not paying for hazmat disposal
>> while shifting costs on to the downstream public in the form of medical bills, cleanup
>> costs, etc.
> I actually agree with this. On the flip side though, is it not just as toxic if we do not ensure that new entrants have at least limited access to IPv4 resources to reach the legacy internet from their IPv6 networks without having to resort to the (out of region) transfer market?

Not at all. New entrants are facing increased costs to communicate with the IPv4 internet, but these are not costs shifted on to them by the malfeasance of others. They are the rising cost of a limited resource running out. Further, these costs are predictable and mostly a one-time expense where the costs being shifted above are unpredictable and recurring until such time as these other entities implement IPv6. 

>> Do you understand now?
>>>>> - As the growing region transits to IPv6 there will still be need for
>>>>> IPv4 meantime. If our IPv4 is not well and meticulously managed during
>>>>> this period it will cost our operators more to buy from the market as
>>>>> AfriNIC runs out completely.
>>>> I actually agree with you here. That’s why I oppose the terrible mismanagement
>>>> proposed in the soft landing BIS proposal which would deprive operators of networks
>>>> in the region of the addresses they need in the present in order to protect imaginary
>>>> future operators who may never materialize.
>>> OK, if you opt to think in binary, you will be right. But you of
>>> course know that this is not how planning and forecasting works. The
>>> region is growing. AfriNIC member numbers are increasing year on year
>>> and most are small players. And when you look at those numbers in each
>>> country and other metrics like internet penetration rates per country
>>> to mention but a few, you know that the forecasts are based on facts.
>>> If dinner was served at your table and you got home before your kids,
>>> would you eat all of it because they are not yet home?
>> That’s not a valid analogy here and you known it. These speculative future
>> startups that are in your forecast aren’t my children. They’re other customers
>> going to the same store that I am going to.
>> Let’s use a better analogy… This is more like a store being operated in a time of shortage.
>> Let’s use eggs for the example.
>> As a store owner, you know that there is a looming shortage of eggs because of some horrible
>> disease that has afflicted all of the local chickens and egg production is less than 1/4 of
>> normal.
>> Would you limit the number of cartons of eggs each customer can buy and prohibit customers from
>> getting in line again if they need more eggs? Would you tell the commercial bakery down the street
>> that you will not sell them 12 dozen eggs because you might have families coming in tomorrow that
>> might need eggs?
>> No, you’re going to pocket the cash as fast as you can and sell the eggs to whoever wants to buy
>> them.
>> Obviously this is still a flawed analogy in that we do actually place some limits and only allow
>> each person to buy the eggs that they can show they actually intend to use and need. (the store
>> owner would not accept or place such restrictions on his commerce)
>> However, it’s at least a little closer. Now, let’s suppose that you do actually tell the baker
>> that he can only have 1 dozen eggs same as everyone else and that he can’t buy any more eggs
>> for 2 years. Let’s assume that these families you expected don’t come for eggs, but are, instead,
>> all lined up outside the bake shop trying to buy bread. Unfortunately, the baker already sold all
>> the bread he could make from a dozen eggs and now he has no more source of revenue. The bake shop
>> closes for lack of revenue and all of those families you thought you were going to help are now
>> going hungry because they cannot get the bread they wanted.
> This is where you start to lose me, because your analogy says that if existing operators don't get the IPv4 resources they require, it will hurt their business. You make no mention of IPv6 in the analogy, and I think that is where the analogy falls flat. 
> Existing operators with IPv4 resources are highly unlikely to go out of business because IPv4 resources ran out. However, a new entrant with only IPv6 space is going to have a tough time retaining customers because they cannot reach the whole internet, and are probably far more likely to fail than the operator who couldnt get more IPv4 resources.   

But that isn't what we are talking about here. What we are talking about is really should an existing provider be forced to pay a higher price for addresses in the transfer market in order to preserve and effectively subsidize possible future competitors ability to obtain addresses at a lower cost?

>>>> Forcing present operators to pay higher rates in the transfer market because they cannot
>>>> get the addresses they need in order to preserve inventory for operators that don’t actually
>>>> exist is nonsensical and quite far from anything I would consider to be “meticulous management”
>>> It is also a bit nonsensical to pretend that there will be no new
>>> operators. This is oblivion at its best. The continent still has many,
>>> many businesses (and other projects such as schools, community
>>> networks etc) that are upcoming within this transition phase that will
>>> NOT afford IPs from the transfer market and they need to be catered
>>> for. Of course an IPv6 only option will not be their solution as you
>>> are aware.
>> I am not pretending there will be no new operators. I am saying that the protection
>> of new operators which may or may not come into being (surely some will, but can you
>> guarantee it will be enough to consume the amount of address space this policy proposes
>> to set aside for them? Didn’t think so) at the expense of existing operators. Let the
>> new operators and the existing operators compete for the addresses on a level first come
>> first served playing field. When the address space is gone, it is gone. C’est la vie.
> I think this is a matter of perspective, but I cannot agree with you here. Requiring a new entrant to potentially shop the transfer market at the get-go to get IPv4 resources to have full internet reachability for their customers significantly raises the barrier to entry for newcomers, and is a sure-fire way to stunt the growth of new operators. And what about non-profit, education and community networks? 

So it is your opinion that we need a "Robin Hood" policy that effectively takes money from existing operators (by prematurely forcing them to the transfer market) in order to subsidize cheaper addresses for possible future competitors. 

I suppose if that is your intent, then supporting this proposal makes sense. However, I consider that to be a form of graft and I cannot support it. 

>>>>> - A mechanism to put in place a carefully managed runout which ensures
>>>>> fair allocation specifically for a region like Africa that has many
>>>>> late business and startups is very critical for us.
>>>> If I were to see such a proposal, I would support it. The proposal that is the subject
>>>> of this thread is pretty far from that.
>>> It does to some extent Owen. It applies limitations to slow down
>>> consumption rates. Your only strong argument is that those operators
>>> are imaginary, which you very well know us a flawed argument because
>>> it defeats the very existence of the concept of planning and
>>> forecasting.
>> But these limitations don’t reduce need, they only reduce consumption. They create an
>> artificial shortage early in order to prolong the duration of the real shortage later.
>> That’s not carefully managing runout, that’s screwing the entire community to protect
>> a small part of the community that doesn’t even exist yet and may never exist.
> Can you explain what you mean here? I don't understand how these are different - surely the nett effect is that there is still a shortage, irrespective of how it was induced. 

Today, there is a free pool and no actual shortage. If we prevent people who need it from obtaining addresses while we still have them, then we have created an artificial shortage at least for those who are denied addresses. 

Yes, there is inevitably a real shortage coming at some point. However, the argument has been made that this proposal postponed shortages. The reality is that it increases the duration of shortages by delaying the end of the free pool and making the effective beginning of shortages much much earlier. 

>> I’m not saying that there will be no new operators. I’m saying that you don’t know how
>> long the space will last under the limitations proposed and that you can only implement
>> those limitations if you prevent people who actually need addresses today from getting
>> them. 
>>>>> - Irrespective and irrelevant of evolution of the proposal and
>>>>> bickering of authors the proposal has the best interests of African
>>>>> network operators and Africa region in general.
>>>> Here we couldn’t disagree more. This proposal has the best interests of imaginary operators
>>>> that don’t even exist and may never exist being placed above the needs of real operators that
>>>> actually have networks and customers they are trying to serve today.
>>>> I don’t deny that the authors genuinely believe that they are acting in the best interests of
>>>> the community. I’m not accusing anyone of malfeasance or malicious action beyond the ad hominem
>>>> and hostile rhetoric which has served only to make it more difficult for the community to find
>>>> common ground.
>>> Yes at this stage let us continue to reason within what is the long
>>> term best interest of our community. In this case I mean Africa. In
>>> the same faith, your analogy of imaginary operators is still baffling
>>> me. I wonder what makes you think the internet in the continent has
>>> stopped growing and that no new operators will emerge.
>> I never said anything of the sort.
>> What I mean by “imaginary operators” is that you are setting aside a vast amount of address space
>> out of the reach of clear and present need today in favor of operators that don’t (yet) exist, with
>> no clear evidence to support any belief that a sufficient number of operators will exist within the
>> useful lifetime of the IPv4 protocol to actually consume that amount of address space before IPv4
>> becomes readily available again due to being generally useless.
>> Protecting the free pool from legitimate use strictly for the sake of procrastinating the official
>> date of runout is like pretending that a bond-issue is free money because it doesn’t immediately
>> raise taxes. I’ve got news for you… Those bonds eventually have to be paid back, so either taxes
>> go up, or, the government cuts services.
>> Depriving today’s providers of addresses deprives real customers of those same addresses today. Is
>> it really so important that whatever small startup may arise 3, 5, 10 years from now be able to get
>> a /24 that it is worth keeping 200+ existing households from getting internet service today?
>> Really?
>> If your answer to that is “no”, then you need to realize that you have been defending a proposal
>> which seeks to do exactly that and consider changing your position.
>> If your answer is “yes”, then likely we have no common ground, my objection is valid, reasonable,
>> and sustained and we should agree that we will continue to disagree.
> Actually, my answer is yes, for two reasons:
> 1) Based on your own defence of IPv6 in this mail trail, why can't these households be serviced using IPv6 at this time? Why defend the use of IPv4 right until there really are no more IPv4 addresses? What exactly is the difference, for an existing operator, between a) no more IPv4 resources due to policy and b) no more IPv4 resources due to there physically not being any more IPv4 resources? They still have no more IPv4 resources either way.

The difference is whether or not it is fair. If everyone faces "no more v4 due to policy" then that's silly, but fair. On the other hand, if you are making operations more expensive for existing providers in order to subsidize their possible future competitors, then that only seems fair if you buy into the whole Robin Hood myth. 

> 2) That /24 could indeed light up an IPv4 internet service for 200+ households, however, it could also provide legacy internet access for many more IPv6 users. Which you think is more important probably depends on your perspective, but I'd rather use that /24 to service a couple of thousand IPv6 users than use it to service < 250 IPv4 users.

This is nonsensical. First, it's just as likely to provide NAT64 to many more IPv6 households at an existing provider as it is with a new entrant. Possibly more likely. There's. Nothing in this proposal to drive the "new entrant" to actually use the addresses in that manner. So the real question is whether it is better to light up however many households today with an established provider or to leave them in the dark until some theoretical new entrant arises to serve them with these subsidized addresses. 

Second, I find the idea that you are using IPv4 to connect IPv6 households to "the internet" to be a reflection of a gross mischaracterization of the terms. There are, in fact, two internets. There is a vibrant and rapidly growing IPv6 internet which is not yet as large as the IPv4 internet, but gaining ground rapidly. Beyond that, there is a haphazardly connected legacy internet full of second class citizens who lack the option of ad hoc end-to-end connectivity due to NAT which has become the hallmark of IPv4.  For this brief period in history, the IPv4 internet is larger than the IPv6 internet. But it already has a much slower growth rate and it's growth is inherently constrained. As such it will be overtaken by the IPv6 internet sooner rather than later. 


> Regards,  
> Andre
>> Owen
>> _______________________________________________
>> RPD mailing list
>> RPD at

More information about the RPD mailing list