After three days of discussions, the sixth Africa Peering and Interconnection Forum (AfPIF) came to a close. This year’s event attracted 232 participants from 57 countries — 40 of which were African countries – and drew 978 online participants from 77 different countries.
The last day focused on “How not to Build an IXP” and the importance of content.
How Not to Build an IXP
For new participants coming from countries without IXPs, the main questions usually are; how much does it cost to set up an IXP? And, should we wait for the local content to grow to set up an IXP or should we set up an IXP and grow the local content exchange after?
The morning session focused on factors to avoid, and factors such as power, and IX sustainability. One compelling presentation demonstrated that it was possible to have a functional IXP on a US$ 1000 budget.
What do you need to do?
1. Market research
2. Build a community
3. Build a technical platform
4. Know who the content providers are in your the country/region
5. Fill up your switch and work on your business plan for the next phase.
Role of Content in Growth of IXPs
The role of content to grow your IXP has become a major topic for AfPIF over the last four years..
Google has been involved in AfPIF since 2010 and is the most common CDN in African countries. Akamai first presented at AfPIF three years ago and has underlined its commitment to spreading its infrastructure in Africa.
In its presentation, Akamai indicated that it was open to serving its CDN traffic, using local caches, to ISP networks through local IXPs. However, the main challenge was the cost of international transit required to populate the local cache, which Akamai leaves to the provider hosting the cache to sort out.
Rwanda,’s IXP – RINEX – explained how they met Akamai during AfPIF 2013 in Morocco, and initiated discussions on the possibility of acquiring a local cache for Rwanda. To meet the cost of the international transit for the cache, the local peers in Rwanda reached a commercial agreement that lowered their per megabit cost for Akamai content. As a result, they were able to increase the amount of local traffic exchanged from a peak of 500Mbps to 1.2Gbps (140%) after delivery of the cache.
Follow-on discussion focused on the issue of locally developed content, locally relevant content, or locally hosted content. It is clear that local hosting is important, for locally relevant content. For instance, major newspapers in Rwanda are hosted abroad, and therefore contributed to the increase in content, with the Akamai cache.
Attracting CDNs to Africa?
Does Africa lack the market to make the business case for CDNs to come to Africa?
Both Akamai and Google noted that with the pace of growth, it was a matter of time before businesses and more locally generated content caught up. They added that even in other markets, it had taken 20 years for the markets to mature.
How do we reduce barriers to content growth?
1. Higher level of trust and collaboration, the rest comes automatically
2. Increase locally relevant content hosting
3. Increase local skills and training
4. Development of data centers and cloud platforms that can host the content
5. Implement policies and regulation that enables competition for terrestrial and cross-border fiber infrastructure.
6. If you don’t have an IXP, then you have no local content to exchange, once you have an IXP, you start thinking of how to improve it.
It is evident from AfPIF 2015 that Africa will achieve 80% local traffic by 2020.
AfPIF 2016 will be held in Tanzania, and we are looking forward to seeing more progress over the next 12 months