IPv4 shutdown in mobile

Mikael Abrahamsson swmike at swm.pp.se
Sat Dec 26 07:38:12 UTC 2015


On Sat, 26 Dec 2015, Mark Tinka wrote:

> None of the major mobile carriers in eastern, western, central and 
> southern Africa have done anything IPv6-related on their network that I 
> am aware about. The availability of IPv4 space in the AFRINIC region, 
> coupled with the ease of spending millions on CG-NAT's vs. rolling out 
> IPv6 means unless they start, someone's pants will be at their knees in 
> a non-picturesque way very soon.

Well, in Europe and Asia, the amount of providers that have rolled out 
IPv6 is not zero, but I'd say it's in the 5-10% range or something like 
that.

It's easier to roll out IPv6 in a mobile network if you can do it single 
stack, so Apple supporting IPv6 only is a major step forward (this is due 
to the IPv6 bearer type was introduced around 12-14 years ago, whereas the 
IPv4v6 bearer type is less than 5 years old and was initially a 4G only 
thing).

> The lack of interest, head-in-the-sand approach is quite alarming,
> particularly as mobile networks are increasingly carrying the majority
> of consumer data traffic in Africa, year-in, year-out.

A lot of providers who have rolled out IPv6 have predominantly done so 
because they happen to have employed good engineers and empowered them to 
do things over longer time, keeping IPv6 costs down because they were done 
during normal hardware/software cycles, instead of a short intensive 
project that cost a lot of money.

I imagine that doing IPv6 single stack rollout in mobile starting now, you 
could be done in 1-2 years without major costs incurred, because now the 
handset landscape has matured enough that there are a good set of devices 
that support (or will support) IPv6 single stack properly. 2-3 years back, 
this just wasn't the case for generic bought-in-the-electronics-store 3GPP 
featurephone or smartphone. It's still the case that if a large part of 
your customer base has simple phones or feature phones, IPv6 support isn't 
needed. So while I do not agree that it's a great business strategy to 
wait even longer, I can understand those who have waited because they saw 
it as too hard to do, potentially because they didn't have the skilled 
engineers to do it. If they start now or during 2016, it's going to be a 
lot easier for them compared to the ones who started in 2010.

I guess there are major differences across the continent as to what 
network gear is used, but I know some operators who shipped their 10 
year old 2G basestations to African providers, and if these are still in 
use, potentially even without a support contract, then these will never 
support IPv6 properly. Networks built like this will be laggards just the 
same way people running old routers won't have the right features to 
support IPv6 properly.

However, if you have 3G basestations with support contracts (or that at 
least have software updated in the last 5-10 years), then getting IPv6 
only working should be perfectly doable if they upgrade the core of their 
mobile network.

-- 
Mikael Abrahamsson    email: swmike at swm.pp.se



More information about the NANOG mailing list