[afnog] IPv6 Progress

Mark Tinka mark.tinka at seacom.mu
Wed Jul 31 21:23:29 UTC 2013


On Wednesday, July 31, 2013 07:51:41 AM Seun Ojedeji wrote:

> > Sure they wouldn’t if it will not bring a few $ out of
> > their budget ;-)

By non-technical users I meant eye balls, not the bean 
counters.

> Yes and No :-) if the provider can still meet up with his
> customer needs(which is largely content related) on v4
> then the pressure for v6 may not be necessarily a
> pressure but a need.

Yes, but your assumption is that the operator is able to 
obtain a pool of public IPv4 space, which is true in Africa 
and Latin America (for now), but less so in other regions, 
particular the Asia Pacific and Europe/Middle East.

Also, content is not as "king" as it used to be. There is 
probably more traffic in Twitter and Facebook than reading 
your daily paper, but this is not an empirical assertion.

> > Yeah because those contents are dual-stacked, how about
> > if they were
> 
> native on v6

I think you mean single-stacked (for IPv6) only.

I don't care if a box is single-stacked or dual-stacked, as 
long as it has IPv6.

Dual-stacking will make sense until new servers can only be 
single-stacked, due to not having anymore IPv4 space.

> > Okay now maybe we use a scenario; here am i in Nigeria
> > sending this mail
> 
> through a mobile network that has over 40million users,
> and YES the network is being NATed. So how much growth
> do you think will be required to move such an operator?

The pressure I'm talking about from user growth is in 
markets that will have run out of IPv4 space.

Operators will NAT and do all sorts of things to keep IPv4 
around, until at some point, it becomes too complicated to 
do so. How long that will take is an exercise left up to the 
readers, but I ultimately think that is what will drive 
adoption.

We are not seeing this in Africa yet because our IPv4 pool 
is still large. It's not enough to NAT, you also need a 
reasonably large public pool to NAT on to, otherwise you end 
up with scaling issues on TCP and UDP port scope per 
address.

> My other question will be; does anyone know a mobile
> operator running public IP to normal mobile users at the
> moment in Africa? if none, what are the reasons why they
> are not deploying v4 especially since its still very
> much available; could one of the reason be that they
> want to leverage on the "security myth" that is
> associated with NAT?

Globally, the "unrestricted" APN provides public IP 
addresses so that users can host services behind their 3G 
connections.

I know Vodacom, Cell-C and MTN have them here in South 
Africa (as do Vodafone in the UK). Can't speak to other 
places.

> If some actually exist(which i am looking forward to
> knowing), why are they not deploying v6; could one of
> the reasons be because their clients are currently happy
> with access to content. Could it also be the actual cost
> of deploying v6 for such a huge mobile network which had
> mostly v4 only devices?

For mobile operators, there are two big issues:

	1. Time to market; with that many users, they always
	   need to deliver solutions NOW. So it's easier for
	   them to justify sending US$40,000,000 on big
 	   LSN's than diverting that money into IPv6
  	   deployment.

	2. IPv6 support on mobile phones is a huge problem.
	   Check these out to see the kind of issues at
	   play:

https://sites.google.com/site/tmoipv6/lg-mytouch
http://wiki.nuevasync.com/wiki/bin/view/Public/deviceIpv6Support

> Well maybe the user growth will help on the long run but
> definitely not in near future.

I'm not looking at a quick win - the long term plan is 
certainly based on user demand in regions that have no more 
IPv4 space.

This is why I focus on NAT64, because it's easy to get rid 
of that without re-engineering the Access.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://afnog.org/pipermail/afnog/attachments/20130731/0a9562a6/attachment.sig>


More information about the afnog mailing list