[Afpif] Transit vs. peering Focust for Africa by 2021

[Mark Tinka]

On 9/Aug/15 11:08, Nishal Goburdhan wrote:
>  
>
> well, sure.  but certainly, in the european context, the production of
> their own local bandwidth (via IXPs) and then content to follow,
> helped tremendously.


Yes, but those exchange points needed a home, which are the data
centres. Then the content owners needed to co-locate as close to the
exchange points as possible to keep their costs down and the end-user's
performance up; which were the data centres again.


>  
>
> simply not true.  one of my favourite pieces of content that’s not
> (yet) localised, is World of Warcraft   ;-)
> ttbomk, that’s not in a carrier neutral data-centre…


This may very well be, but the problem with facilities that are not
carrier-neutral enforce limits on who can enter that facility. Or if
they allow 3rd party networks in, enforce strange tactics such as
over-priced x-connects, "guidelines" on whom you can interconnect to,
what you can bring in, how much of it you can bring in, access to your
rack(s), e.t.c.

Almost every major service provider has their own facilities within
their countries. It is not uncommon for their customers (like World Of
Warcraft, for example) to co-locate with them if they primarily buy
their IP services. Now, if another network wants to connect to World Of
Warcraft in that data centre, there aren't guarantees that this will
happen easily, quickly or at all. And that is the issue with private
facilities. They are an essential part to growing the eco system, but we
need to go one step further to truly reach the promised land.

And to top it off, some operators in Africa still mistrust one another,
and will not co-locate their infrastructure in their competitor's
facilities.


>
> to be clear, i’m not disagreeing with you about the data-centres.  i’m
> saying that instead of looking at the “100% completed product” do what
> you can now, with what you have, and work towards that.  even a
> “non-carrier-neutral” data-centre that allows you to host content
> locally, is a win.  and the misinformed, that religiously insist on
> this as a pre-requisite to doing business locally, are only shooting
> themselves in the foot.
> i think back to when i was an operator;  we ran many successful, and
> *full* data-centres that while not neutral in nature, certainly added
> value to our local ecosystem.


I do not disagree with you there, and this is already happening all
across Africa.

The question was whether we can switch the 10:90 (Africa:Non-Africa)
traffic ratio now to 80:20 by 2021. To do this, we will need a major
shift in how we handle co-location on the continent. Regardless of that,
our long term goal remains.

Agree that there are several non-African networks that have ignored (and
continue to ignore) Africa because of the lack of carrier-neutral space
for their requirements. I applaud those that have had the courage to
come down and use whatever little resources we have. It will certainly
give them the leg-up when the rest realize that they should have been
here sooner.


>
>  
>
> i’d disagree.  it MUST (2119-style) be a long term requirement.  how
> else are we going to hope to fix the current “internet transit
> deficit”, to borrow a phrase from afpif-3 ?


I'm not saying we should not localize content.

I am saying that to hit the 80:20 ratio by 2021, it is not the panacea.

Of course, I fully support localization of content. That is, perhaps,
the only way we shall increase penetration as not all of Africa's
populace are English-speaking.


>
>
> YES!  so, let me ask you this?  are you saying that the real failure
> here is an inability to develop interesting local content, regardless
> of where its hosted?  :-)


No.

I'm saying that to reverse the traffic ratios to 80:20 in favour of
Africa, we will need the localize at least 90% of that traffic within
Africa.

Yes, a lot of content, software and applications have been developed for
Africa by Africans, and I continue to support and push for this in order
to increase penetration for non-English-speaking folk. However, a
greater portion of the 90% of traffic Africa is pushing to Europe is
from some big global names. The Internet being what it is, African
content developers are going to have to compete with the rest of the
world for a piece of the African eyeball pie. Can't hide from that...


>
>
>
> that’s certainly the important question for the operators who run the
> Fat Pipes that carry this content now.
> but i think michuki is hinting at what people see as a digital future
> for africa.   or shall we continue to remain an exporter of just
> bananas, and not digital content .. ?

Well, I was focused on the question at hand, which is how we shift 80%
of traffic from Europe to Africa by 2021. The short to medium term
solution is to localize that popular global content. The long term is to
develop localized content in order to increase penetration. In parallel,
develop localized content that can compete with the popular global
content. Users don't really care about what you, I and the rest of the
Internet operator community in Africa are trying to do for Africa. As
Randy says, "They just want their MTV". It is up to African developers
to have a version of "MTV" that will rival the incumbent one. No way
around that, unfortunately.

However, that does not change the fact if we want 80% of Africa's
traffic remaining on-continent by 2021, we need to move the popular
traffic our users crave from Europe to Africa. What is needed to do that
is what I've already argued.

Mark.