If peering is so compelling, why isn’t every network full-mesh peered? Wouldn’t that lead to a transit free Internet?
What is the native language spoken between peers?
How many IXPs are there across Africa and how many operate route servers?
Peering provides access to each others customers, but there tends to always be a need to get traffic elsewhere. For example, an Internet Service Provider that peers its traffic at the local Internet Exchange Point still needs to also reach the rest of the Internet. Every ISP by definition sells access to the entire Internet. The ISP could extend its reach to all Internet Exchange Points all over the world, but this would be at great cost, and more importantly, not all network operators will peer with them. As a result of this incompleteness, ISPs generally need to purchase transit from someone to reach some destinations on the Internet.
The short answer is English, and BGP, depending on whether you are a carbon-based life form, or a silicon life form....but I think you are bringing up the Language Barrier as an obstacle to peering.
At the African Peering and Interconnect Forum (AfPIF) in Casablanca last week we witnessed the Language Barrier first hand. I ran the Peering Simulation Game, and two of the players spoke English pretty well but their primary language was French. And during the game it became clear that the language barrier led these players to not understand the rules and negotiations of the game. Some of the moves demonstrated this as well, but irrational moves also occur with native English speakers; in these cases it is the misinterpretation of the rules that causes the problem. The negotiations at AfPIF brought forward some misunderstandings perhaps based on some nuances in the language, and then perhaps the resulting unfamiliarity with the terminology. All of this led to peering negotiations that simply didn’t make sense.
And so, I left feeling that the Peering Simulation Game was a failure. The players struggled, and the audience seemed a bit restless observing these struggles.
However, the feedback after the event was entirely positive!
The audience members I spoke with said that they observed a live demonstration of the challenges with peering across the language barrier! This is one of the challenges that they face across Africa, and we brought it front and center, demonstrating the challenge in our live simulation for all to see!
I know that I have seen the language barrier in real life presenting an obstacle for networks seeking to peer. Many times it is poor grammar and misspellings in the peering requests that imply a language barrier or a lack of care might present problems when establishing the peering session as well. It may therefore be difficult to establish rapport with the peer, and agree upon the mutual benefits of peering.
These difficulties occur after the first challenges in finding the right person to speak with, and knowing the right information to communicate to maximize the chances of obtaining peering. Too much information or too little information in an email leads to the peering request email being dropped. The language difficulties amplify the perception that the peering may be more trouble than it is worth.
The greatest obstacles to peering are not technical.
I believe Euro-IX has one of the best lists of African IXPs:
Wikipedia has a list as well, recently modified:
As you know, the Internet Society has been doing great work promoting the construction and development of IXPs across Africa through training. You can read more here: http://www.internetsociety.org/what-we-do/issues/internet-exchange-points-ixps
And here is a graphical representation of the PCH IXP directory on IXPs from the NSRC. Click on the map and it will take you to the list of IXPs.
As for whether they have route servers or not, the IXPs will often advertise route servers on their web pages. I would expect there would be many route servers across Africa because the technology has hardened, it is free, and eases establishing many smallish peering sessions.