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

[Meoli Kashorda]

Dear Mark,

Thanks for your very detailed responses. This is excellent preparation for AFPIF 2015.

On the 60% local content, I am the one who misread the note from Michuki (have not read the actual report). It suggests we shall achieve 60% not 80% by 2021.

I have no data to challenge their forecast. But I do have some anecdotal data on Kenya -  global CDNs are saving Kenyan operators about 50% of international circuits capacity! And that Google and Akamai caches are serving about 35-40% of traffic from Kenya. So we have probably achieved 40% local traffic? 

Maybe someone has already done a country by country study? We at KENET would appreciate a research grant to conduct EA study - anybody generous? 

I wonder why most of us are so quiet - maybe we already have anecdotal data on each country represented in this mailing list? And Michuki could compile it for AFPIF 2015 in two weeks? 

Meoli





> On Aug 10, 2015, at 7:55 AM, Mark Tinka <mark.tinka at seacom.mu> wrote:
> 
> 
> 
>> On 9/Aug/15 21:01, Meoli Kashorda wrote:
>> 1. First the original post by Michuki suggested that data was showing that we had already achieved 60% content served from Africa! So we had only 20% to achieve 80% by 2020. It is not  20% local and 80% global but 60% local and 40% global! This is progress.
> 
> 
> I don't think the report said we are 60% today. Unless I mis-read it,
> the report says that peering traffic will grow by 60% between 2014 -
> 2021, putting Transit traffic at 40%.
> 
> To be fair, it is not clear whether 60% of peered traffic by African
> operators will happen in Africa or outside Africa.
> 
> I suppose we can infer that peered traffic takes care of Africa's demand
> for global content. What we don't know is whether this peering is
> happening in Africa or not.
> 
> If you ask me, I do not think we are keeping 60% of Africa's traffic
> demands in-continent. Last time I did a rough poll of my trends, we're
> sitting at about 92% of traffic demand being served from outside of
> Africa. Long way to go still.
> 
> 
>> 
>> 2. Sustainability or profitability of African operators building fiber and 3G /4G national networks is significantly enhanced by presence of Global content distributors (Google and Akamai in Kenya) having a presence in our countries (not continent because cross-border leased lines are still very high, at least in EA).
> 
> 
> The strangeness of it all is that trans-continental circuits are cheaper
> than national or regional ones.
> 
> This phenomenon is not unique to Africa. You find it in developed
> countries as well. My summary, "the shorter the line, the dearer you pay".
> 
> 
>> 
>> 3. Both Google and Akamai do not seem to care too much about carrier neutral data centers. Just data centers, even Tier 2 data centers. Our challenge is the high cost of broadband leased lines used to distribute broadband content not collocation or even the problems lack of carrier-neutral data centers.
> 
> 
> Not quite true - both Google and Akamai (like most other content
> players) do care about co-lo.
> 
> The GGC (Google Global Cache) and Akamai's clusters, that typically get
> deployed within a service provider's network are meant to serve the
> eyeballs that service provider has. In some cases, that service provider
> will work with Google and Akamai to share those caches with other
> service providers in the country via private links and/or an exchange point.
> 
> These caches are designed to be deployed in this fashion, because the
> scope of services delivered over them is not one that would require
> extremely stringent co-lo properties. If the power failed, for instance,
> traffic will be re-routed to the next nearest cache; the only side
> effect being an increase in latency.
> 
> Now, if Google want to deliver their cloud applications (Google Docs,
> Storage, Compute, Big Data, e.t.c.), they will be looking for data
> centres that meet their minimum stringent requirements, and also present
> a decent opportunity for peering with a national exchange point. Those
> kinds of deployment would be few and far between, in Africa.
> 
> Similarly with Akamai, the story is very different if they are deploying
> clusters that are going to deliver HTTPS content.
> 
> It's all about the use-case.
> 
> All that said, however, I have to applaud all of the content players who
> have all deployed their CDN platforms in Africa. They see the
> opportunity, and their "braveness" will pay off.
> 
> 
>> 
>> 4. I think local content distribution will be subsidized by the global content in Africa. In any case, what is local content? The local TV stations all have YouTube channel and even universities have a YouTube channel and that is what is what is driving Google traffic. Google Mail is carrying local email. Facebook has a lot of what is really local content that users want to access. How will you measure local content on a global CDN?
> 
> 
> And this is a very good point.
> 
> Remember the Internet is borderless. When Africa creates more content
> and applications that gain global recognition, demand for that traffic
> is going to be global. Who do you think is going to deliver that
> traffic? You guessed it, the global CDN's like Akamai, Limelight,
> EdgeCast, CDNetworks, e.t.c.
> 
> By the same token, as Africa continues to digitize, we will want to
> further optimize our own African content for consumption within Africa.
> Who do you think will deliver content built in Somalia to eyeballs in
> Namibia. You guessed it, the global CDN's, or even an African CDN player
> who comes up to compete with the global CDN's in Africa.
> 
> This is why I continue to say that rather than focusing on generating
> African content to solve our traffic ratio problem, Africa should pay
> close attention to a) getting the global players on-continent, and b)
> competing with the global players in Africa at least, and globally next.
> But at the same time, continue to develop local content in order to
> increase penetration for those that do not speak/understand English.
> Higher penetration gives us leverage, which generates both local and
> global interest.
> 
> Mark.
>