Thanks for the great info. Having talked of onward connectivity dont you think it would make enormous sense if say SEACOM and EASSy was to link up with WACS, SAFE or MAINONE around South Africa? In the same regard there is already talk of TEAMS and SEACOM linking up through KIXP(Kenya Internet Exchange point) or via Mombasa. Infact if all the cables on the East African coast materialize then Kenya will indeed be a hub as will South Africa because many landlocked countries vizualize Rwanda, Uganda, Burundi, Ethiopia and South Sudan will want connection to the cables which all have landing stations in Kenya. Given the BPO activity and the development of local content we could begin to create a serious broadband traffic node with connection to the Middle East, India and Far East in addition to the traditional London. In addition the numerous terrestrial national fibre-optic networks being rolled out all over Africa will mean greater African inter-connectivity which then reinforces the ring around Africa concept. Your thoughts on this?

On submarine fibre you buy connectivity from one landing station to another. The amount of connectivity you buy is calculated either in STM1′s or in wavelengths. Arriving in a landing station is by itself not enough to get connectivity to the rest of the world. To get connectivity to the rest of the world, a network needs to buy onwards capacity to other locations or if it wants “generic” internet, it needs to buy transit and get peering arrangements.

Whether it wants to buy generic internet or dedicated onwards capacity to another location is very much dependent upon where the line terminates and what kind of customers the network has. Say it is a large telco and serves a host of banks and mobile networks with offices in Africa, the Middle East and India. These demand short roundtrip times, guaranteed bandwidth, automatic failover of traffic in case some fisherman breaks the fiber etc. In that case it wants to deliver guaranteed bandwidth between Mombassah, Jeddah, Capetown, Mumbai etc. It will therefore want to be connected to TEAMS or Seacom and buy onwards connectivity to where the customer is located.

If it is a large African ISP however, it might figure that what it really needs is cheap transit and lots of peerings on a global scale. In that case it might just hook up to one of the fibres going to Europe, buy onward connectivity to a couple of the internet exchanges in Europe and try to connect either privately or over an internet exchange with as many networks as possible and buy as much transit as it can for 4 euro/mbit/s/month as it can. This might not always give the shortest round-trip times to the world east of Africa, but it would be cheap. And chances are that it will get better interconnections/transit deals with the likes of Flag or Tata in Amsterdam or London, than it would in Africa, because the Tata’s sales organisation in Europe couldn’t be bothered by the loss of revenue in Africa. And if Tata isn’t willing to deal in Europe, there is always someone else who’s willing and able to make a special price, cheap cheap.

This is why Seacom have negotiated transit to London even though their cable lands in Marseille because London is much more of an Internet hub than Marseille, although why London and not Amsterdam or another European hub, I have no idea.