I spend most of my time teaching engineers in different countries how to plan and deploy IPv6 networks. Over the last two years, I have been speaking more and more to non-engineers. These are either technology executives who sense that they need to do something about this “IPv6” thing, or government IT leaders who want to understand what the problem is and more importantly, what they could do. The most impactful part of these these exchanges is when I get these managers to understand the implications of IPv4 address exhaustion to their organisations. As I work for a Regional Internet Registry (RIR and no, this post is not the official position of my employer), I thought I’d put down some words on the implications of Africa being the last continent to run out of IPv4 address space.

To be clear, by IPv4 exhaustion, I mean the fact that there are no more IPv4 addresses available in the central pool of addresses managed by IANA to hand out to different RIRs as has been the case. This was officially announced on the 3rd of February 2011. Thereafter, all the various RIRs have been burning through their last blocks of IPv4 addresses and are all (with the exception of AFRINIC) on IPv4 address life-support.

Implication #1: 10 Africans Per IPv4 Address… Forever

Approximate distributions of IPv4 address per capita by continent

Post-exhaustion, the total number of IPv4 addresses assigned to Africa is 84 million (i.e. 5 /8 blocks; each /8 being a block of about 17 million addresses). With a population of about 1.2 billion, we have a per-capita IPv4 address of 0.07 for the entire continent. Rounded up, it means that ten (10) Africans still have to share one IPv4 address. Drill down deeper and the situation is a gamut that runs from 2 people per IP address in South Africa to more than 1500 per IP address in Eritrea. Because the global pool of IPv4 is finished, this dubious fate is not going to change so long as we continue playing in the bloody red oceans of IPv4. Coming late to the Internet party was bad enough, getting stuck in a disadvantageous situation with no hope of change is terrible. Fortunately it is not a hopeless situation, hope just takes the form of the bountiful blue ocean of IPv6.

Implication #2: Not Just Land-Grabs… Welcome IPv4 Address Grabs

Available /8s in each RIR as at 15th March 2015 courtesy NRO.net

Consider the following:

• Research indicates that doubling Internet broadband speed results in 0.3% increase in GDP which is about $126 billion for OECD countries (based on a study of 33 OECD countries by Ericsson, Arthur D. Little & Chalmers University of Technology)
• There’s been about an 80 fold increase in number of Internet users in Africa between 2000 and 2014. Primarily driven by mobile, fixed broadband is also on the rise.
• As broadband becomes more prevalent, the bottleneck of Internet performance will shift from poor bandwidth to things like NAT gateways which are heavily deployed on the continent.
• Africa is already disadvantaged with respect to IP addresses on which the traditional (hopefully soon to be historical) Internet runs.

It is obvious (at least to a smart engineer) that one of the keys to maximising the potential of the broadband infrastructure deployments is to address the bottlenecks created by NAT devices by getting rid of them. Getting rid of NAT means using more unique public IP addresses wherever possible. As such, you could be forgiven for thinking that Africa’s consumption of IPv4 addresses mirrors the Internet growth and that all of those 84 million addresses available to the continent have been snapped up for use. You would be dead wrong! About 55% of those 84 million addresses are sitting unallocated at the AFRINIC Ltd (the RIR for Africa).

To put it in perspective: Africa still currently has the largest unused share of a resource that is

a) valuable (worth $7 – $12 each or about $500 million by a conservative estimate)

b) in high demand in other parts of the world (the reason doesn’t matter)

c) currently powers about 90% of today’s Internet traffic

What could possible go wrong given the resulting dubious economic incentive? At least one thing I can bet on; greed (aka human nature) will respond to that incentive. International big business has often come to Africa (and in collusion with some Africans) to reap (even rip); from slaves, oil, gold, diamonds, copper to land. Now we can add the already small pool of IPv4 addresses to the mix of resources that these organisations are coming for. Expect to see one or more of the following occurrences:

a) Shell companies being set up in African countries where it’s “easy” to do business (arrive on a plane in the morning and by evening have a fully registered company) so they can qualify to request for some of these addresses from AFRINIC.

b) African subsidiaries of global multinationals start requesting larger amounts of address space than they might need which could possibly be re-allocated internally (to the company) and off of the continent.

c) Legitimate African countries responding to the incentives presented by an IPv4 Address market start over-stating their requests to AFRINIC, getting them and selling off the extra addresses on the IPv4 transfer market.

An effect of this will be that the folks at AFRINIC who evaluate IP address resources will become more thorough (and probably take much more time) in assessing requests as the try to determine legitimate requesters from fake ones.

We’ve seen people declare that if they cannot get IP addresses, they will not invest in the Internet infrastructure in Africa. (and indicative of the kind of investor we don’t want): the economic opportunity in Africa’s Internet landscape is due to the high number of people who still don’t have reliable broadband (read NATless) Internet. There’s never going to be enough IPv4 to achieve that and so that’s going to have to be delivered over IPv6. Yes, while a pool of IPv4 addresses is absolutely critical in the short to medium term for these projects, there must also be balanced approach to ensuring that a few organisations don’t take an unfair share of what is a critical resource that a lot of other operators are going to need to facilitate their transition to IPv6.

Implication #3: Driving Complacency in Africa Towards IPv6 Adoption

So long as it is still possible and easy to get IPv4 addresses, 95% of network operators will do so; humans (and the organisations that they make up) don’t change easily. Particularly, organisations fixated on short term results will fail to realise that using every every single IPv4 address available to AFRINIC is not enough for us to catch up with the rest of the world. Not being able to look beyond a the proverbial tree to the forest, it is easy to totally be blind to drastic change in resource availability that will result if one of Africa’s large mobile telecommunications companies (say MTN) decides to remove NAT on their mobile networks and give each mobile user a true Internet access experience using public IP addresses (I didn’t think it was possible until experienced it for myself in Dar es Salaam last year with Airtel).

Networks will move to deploy IPv6 under the following influences:

a) The cost of getting IPv4 addresses becomes un-bearable (as would be if they have to be acquired from the Transfer market)

b) There are no IPv4 addresses to be gotten, anywhere (even if you have money to buy them)

c) The network’s owners have this blinding insight of the obvious that the IPv4 game is an un-winable one and IPv6 is the only sustainable way

d) Some powerful entity forces them (by mandates or by incentives) to deploy IPv6 (like gov’t IT regulators)

For most African companies, scenarios (a & b) are unlikely to happen in next 2 – 3 years (unless the shell companies increase and succeed in getting addresses which depletes AFRINIC’s current space). Only a few network operators to my knowledge pay anything other that lip service to scenario (c) and the only party that can make scenario (d) happen (gov’t regulators) still largely don’t know how to go about it (if you are an African gov’t IT leader and need help with this, I might be able to help). Therefore, we have ourselves an IPv4 address resource curse!

Implication #4: Africa Becomes a Dumping Ground for Legacy Technology

Source: http://cdn.c.photoshelter.com/img-get/I0000zm8Zl7KLeDU/s/860/860/Agbogbloshie19.jpg

“Wherever there is money to be made, it will be made—damn the ethics of it, especially if there’s no watchdog” —Mukom Akong

Consider the following events that are either already happening or could happen within the next 3 years.

1. In response to growing Internet demand worldwide, vendors increase their inventory networking devices and software.
2. Since not many buyers know enough to insist on IPv6 capability, vendors either don’t care about it (my home 2014 model 4G LTE modem from a popular vendor doesn’t even support IPv6. Ironically, the operator who provides this service did a big show of “launching” IPv6 last year) or pay lip service to it with shoddy and incomplete IPv6 implementations.
3. Complete IPv4 exhaustion at the RIR level which further raises the cost of acquiring IPv4 addresses.
4. Many big network operators take the leap and start deploying and using IPv6 as primary protocol (e.g. T-mobile in the US has about 40% of its traffic on IPv6).

What happens to the large inventory of IPv4-only equipment when the large networks in developed economies start insisting on IPv6 support in all their purchases? Combine that with the complacence on the part of African network operators, weak IT regulatory regimes in Africa and the result is Africa becoming a dumping ground for all of that equipment (as the vendors scramble to off-load these toxic assets)—either as cheap equipment for the market or as “technical aid”. Such a glut of legacy devices will sabotage Africa’s ability to deploy IPv6. Following the legacy devices will be a plethora of legacy IPv4-support services that don’t serve the long term technical interests of the continent e.g. NAT-optimisation devices and consultants. Of course the vendors get to empty their warehouses of legacy equipment while either making money or reducing their financial burden of having to get rid of it in environmentally friendly ways in strict environmental regimes. Heck, they might even score points with less savvy audiences for doing “aid” or technology transfer. I’ve seen this before—stores filled with ‘donated’ networking equipment that the recipient was unable to use because the technology was old and worthless or never even existed in most African countries (think hubs and routers with ATM interfaces in Northern Cameroon or Nigeria). Yes I am suggesting that when it comes to technology, we should disobey our parents’ advice and “look the gift horse in the mouth.” If the horse can only galop in 32 and not 128 units, say “Thanks but No”

Implication #5: the Second Digital Divide, With Africa on the Wrong Side… Again!

Earth at Night, courtesy of NASA

Either out of necessity or because of vision, the majority of the world will go mostly IPv6. I suspect that the US market alone at some point will tip the scales and the rest of the world will just follow. If either due to a lack of vision and or complacence, African networks remain mostly IPv4, the result can only be another bad prophecy that we’ve seen in various pictures like this one above; the Second Digital Divide, with Africa once again on the wrong side.

In the short to medium term, networks run by operators who understand that the future is IPv6 will run both IPv4 and IPv6. However at some point, it will not just be feasible to run dual stack any more due to a lack of IPv4 addresses. Meanwhile some of the remaining rough edges of IPv6 will have been ironed out through experience and incremental updates. When the costs and complexity of running both IPv4 and IPv6 outweigh the benefits to maintain connectivity for legacy (IPv4) traffic, such operators will drop IPv4 and run only IPv6 . At that point, in order to still remain connected to the mostly IPv6 Internet, African networks will have to engage in hasty, and expensive big bang IPv6 implementations or worse invest in black boxes to interconnect. Either way, it will be costly price-wise (in the case of the former) and performance (in the case of the later) which does’t bode well for African Internet consumers nor our digital economies. Think digital isolation for some.

Africa must use it’s current pool of IPv4 addresses as key strategic asset in migrating to IPv6 as soon as possible. We have the benefit of not having lots of legacy network technology and quite many greenfield Internet deployments. Making all these huge investments and not future-proofing them by thinking IPv6 first with support for IPv4 in the short to medium term is beyond that stupid, it’s economically and intellectually criminal.