[AFRINIC-rpd] Academic IPv4 Allocation Policy Second Draft (AFPUB-2013-GEN-001-DRAFT-02)

[Andrew Alston]

HI SM,

> I went through some old statistics.  Internet access in some parts of Africa when compared to the U.S was 100 times more expensive.  With the necessary infrastructure Internet access is twice as expensive when compared to the U.S.  The following was said four years ago:
> 
>  "IXPs in Africa count their throughput in megabit per second, and sometimes even
>   kilobit per second. The rest of the world tends to be concerned with gigabit
>   per second."
> 
> There hasn't been much of a change since then.  There is a well-known provider established outside the region which sees about 50% of its traffic coming from a well-known company.  Some parts of that traffic ends up in Africa.  Anyone who has an interest in Africa as a whole might wish to look into the details and see whether it can help reduce the cost of Internet access in the region.
> 
> Getting IP addresses cheaply is a minor part as the picture.  If this proposal can ease the distribution of IP addresses to over 50 universities in the region (I excluded Nigeria and South Africa for obvious reasons) it can be a good thing.
> 
> From an IP addressing perspective the following suggestion is a bad idea.  I suggest adding a condition in the proposal for connectivity to an IXP within the region.
> 

Hi SM,

Firstly for the purposes of this list and this mail, I write this one in my personal capacity and what I say here does not necessarily directly reflect the views of the UbuntuNet Alliance or any other network (sorry to start with a disclaimer, but I prefer to keep the boundaries clear considering my involvement with many projects)

This policy deals with providing end user space directly to institutions, they may or may not choose to be directly at IXP's, that being said.  Most universities who fall under this policy will have connectivity via their NREN's, and in turn, especially as time moves forward, those NREN's will have a connection to their regional academic network (UbuntuNet in Southern/Eastern Africa, WACREN in Western Africa and ASREN in the Northern areas).  The regional networks I believe (and I can say for sure this is the case with UbuntuNet) aim to be at every IXP they can whereever feasible in the countries they have points of presence.

The cost of infrastructure and bandwidth will drop drastically with the construction of certain networks in the coming the months for academia as well (The AfricaConnect network being one such network, that should connect South Africa -> Mozambique -> Tanzania -> Kenya -> Uganda -> Rwanda, I suspect during Q2 this year, and further countries will be added to that list in subsequent phases).  You can read more about this one at www.ubuntunet.net / www.africaconnect.eu.

It is pointless forcing academic institutions themselves to go to IXPs though, because as I said, either their national research and education network will have presence at such, or alternatively the regional academic network will.  The NREN's at IXP model is already active in Africa, with South Africa and Kenya being the prime example of this (KENET, the Kenyan National academic network) if I recall is the largest single traffic destination on the KIXP, CINX was re-established largely on the back of TENET's (the South African NREN) presence there and willingness to feed content.

I can fairly safely say on this list though, the issue of getting nren's and ren's to IXP's is a very high priority topic and is being worked on.  After all, one of the stated goals of the Africa Connect project is to keep African academic traffic within Africa, and IXP's play a pretty crucial role in that.

Andrew