[AfriNIC-rpd] Updated Version of the "IPv4 Soft Landing Policy"now Available Online

[Owen DeLong]

On Feb 24, 2011, at 6:20 PM, David Conrad wrote:

> On Feb 24, 2011, at 1:03 PM, Owen DeLong wrote:
>>> I think the point Andrew is trying to make is that we are NOT all in the same boat.  There are 5 different boats, some of which are moving much more quickly to the falls than others.  As a result, attempts to come up with pieces of paper that deny folks from trying to get onto the slower moving boats aren't likely to be effective and instead, will simply result in folks sneaking on board the slower boats.
>> Why is it better to allow them to overload and sink the boat
> 
> I believe you missed my point. Instead of coming up with pieces of paper folks will simply ignore, an alternative idea is to have the boats go over the falls at roughly the same time to reduce the incentive for folks to try to sneak aboard the slowest boat.
> 
Oh, I see... Instead of trying to keep the slow boat from sinking before it reaches the
falls, you want to step on the gas in order to rush over the falls at the same time as
the faster boats.

I don't know about you, but, when I see cars in front of me about to collide, I do not
tend to step on the gas to see if I can join the accident.

>>>> The difference is that depleting AfriNIC prematurely by exporting resources to other regions makes the transition that much harder in the AfriNIC service region.
>>> Sorry, how is transition made harder?
>> If you don't think that a certain amount of IPv4 space is necessary to virtually any viable IPv6 transition strategy, then, you aren't paying attention. If you do, the, you should readily recognize that a lack of IPv4 space with which to deploy your transition strategy will present an additional challenge.
> 
> Sorry, still not following, so I guess I'm not "paying attention".
> 
> According to http://www.potaroo.net/tools/ipv4/rir.jpg (and assuming the laws of economics are suspended :-)), AfriNIC will be allocating IPv4 until somewhere around 2015.  This means that folks in the AfriNIC region will still be deploying IPv4 _four years_ after folks in the APNIC region have run out of IPv4 and are (hopefully) deploying IPv6-only.  You claim that this will make transition easier. Seems to me it would be encouraging folks in the AfriNIC region to delay the transition, thereby making the ultimate transition harder (since there would be more infrastructure deployed on IPv4 that would need to be transitioned).
> 
I think Geoff's numbers are optimistic at best. However, even if AfriNIC runs out four years later, the amount
of off-continent communication required in the AfriNIC region will create a demand for dual-stack within
the AfriNIC region well before AfriNIC runout.

I think that transitioning before runout in the region is actually demonstrably easier than transitioning
after runout. Of course, if allowing runout in the region to happen later leads to procrastination, then,
there is, indeed, a risk of self-destructive behavior in the region causing damage and additional
hardship as it has in the other regions.

I guess I'm hoping that, as with the cellular vs. wireline telephone deployment, Africa will be able to
take advantage of lessons learned the hard way in the rest of the world. (Yes, I realize you
attempted to apply this idea in the opposite direction earlier, but, I think it fits better here, so,
I'm using it anyway.)

>> First, I think that much of the content consumed within the AfriNIC service region is not hosted within the region.
> ...
>> Second, much of the content within the AfriNIC service region is intended for external consumption such as tourists and attempts to attract other international business.
> 
> I'll admit to not being convinced of your assumptions regarding the use of the Internet in the AfriNIC region (I actually suspect most traffic in the AfriNIC region is/will be produced and consumed in the AfriNIC region), however ignoring that, according to http://www.potaroo.net/tools/ipv4/rir.jpg sometime before 2013, folks in the APNIC, RIPE, and ARIN regions will have exhausted their remaining IPv4 free pools. Pragmatically speaking, this will mean folks everywhere will (we hope) need to be dealing with IPv6-only. Of what benefit would deploying new infrastructure based on IPv4 have in this situation?
> 
I would presume that new infrastructure would be deployed based on dual-stack, not IPv4-only. The question is
whether AfriNIC is better in the 2013-2015 being able to deploy native dual-stack, or, having to deal with
all IPv4 needs through multi-layer NAT and other hackery.

Given that there seems to be near universal agreement that such IPv4 hackery is hideous and will generate
a relatively poor user experience at best, I'm inclined to believe that reducing the probability of a need for
such is of benefit to the community.

Perhaps you are correct. Perhaps it's best if we break everyone's internet at the same time. I think there
is some potential advantage to the AfriNIC community if they break later, but, if they want to join the
rest of the world and race over the cliff at the same time, they are most welcome.

Certainly I think that the AfriNIC community should make a conscious decision between these
alternatives. While I will express and advocate my opinion that the region is better off preserving
IPv4 resources for this purpose in region, I will not be voting on the matter and firmly believe that
the consensus will of the community within the region should be what is ultimately implemented.

Owen