A LACNIC Initiative in Africa
The Simón project, a LACNIC initiative for measuring Internet connectivity in the region, conducted research on network interconnection in African countries.
Agustín Formoso, a software developer with LACNIC’s Technology department, was responsible for this project and presented the results of the study at the AFRINIC 25 meeting held in Mauritius (https://www.youtube.com/watch?v=uxGKbAOwSkM )
In an interview with LACNIC News, Formoso noted that the information gathered during the research proves that African networks share similarities with those in Latin America and the Caribbean, such as the fact that traffic is exchanged outside the continent.
He also observed that the results make it possible to compare the relative connectivity densities in both continents.
What is the Simón project? How did LACNIC come up with the idea for the project?
The Simón project offers up-to-date and representative information on connectivity measurements in Latin American and the Caribbean (LAC). It originated around 2009 during regional talks on interconnection. At this moment the project is growing and in search of volunteers who wish to collaborate.
How can we collaborate with the project?
How do you measure latency between networks?
This can be done in several ways. In this study, Internet Control Message Protocol (ICMP) packets are sent from various origins to various destinations across the region and the time it takes to obtain a response is measured. The project also has a web meter that allows us to trigger HTTP measurements from a web browser. ICMP measurements and web measurements provide two different perspectives of the network: both are located in different logical layers, but both reflect how the network is viewed by the end user, as all measurements originate from average user devices.
Why is it useful to measure latency?
Latency is a good indicator of the efficiency of packet routing between source and destination. Good latency values correspond to good interconnection between networks. Good interconnection means that traffic uses the network resources it needs, not more than it needs (as is sometimes the case). It means that local traffic remains local. For Internet users, this translates into better quality of the services they access. While there are many other sources that generate delay, we believe that sub-optimal routing is one of the major contributing factors at network level.
Can latency measurements help improve Internet connectivity?
Yes, they can. Before the Internet can improve as a whole, we first need to improve the networks it comprises. The results of this study are of particular interest to network operators, as they can find and download the results that involve their own networks. If operators can identify sub-optimal connections and use this information to improve their connectivity to other countries or operators, the Internet as a whole will improve.
Why measure latency in the AFRINIC countries?
Taking advantage of the collaboration with our peers on the AFRINIC research team, we wondered why not extrapolate the measurements to the African region? African networks share certain similarities with those in our region, such as the fact that traffic is exchanged outside the continent On the other hand, they must deal with the disadvantages that are typical of the continent where they are located: its geography, cultural diversity, and economy. Conducting the study in Africa allows us not only to analyze the African continent, but also to have a reference level against which to compare our own region. Further details of the study can be found at LACNIC Labs http://labs.lacnic.net/site/connectivity-in-africa.
According to your research, what can you tell us about the quality of the Internet in the AFRINIC countries where latency was measured?
Connectivity in Africa turned out to have two very different components: quite a good one in the north of the region (near Europe) and a very bad one, particularly in the central part of the continent. These results strengthen the hypothesis that being located near a traffic exchange region improves connectivity towards our own region. In this case, the African countries closer to Europe proved to be better connected with Africa itself. This highlights the importance of keeping local traffic local and the relevance of implementing and using IXPs in the region.
Is it possible to compare the results of latency measurements in the LACNIC region against those of the AFRINIC region?
Yes, they can be compared, as both regions used exactly the same methodology for the study (http://labs.lacnic.net/site/connectivity-in-the-lac-region).
At first glance, a comparison of the two regions appears to show that Latin America and the Caribbean have better latency results. The comparison requires taking into account the geography of each region: while there is a high density of countries in Central America and the Caribbean and this tends to favor latency measurements, Africa is made up by large countries evenly distributed throughout the continent, making it more difficult to achieve connectivity levels similar to those in the LAC region.
On the other hand, both regions showed similarities and differences in terms of the results obtained in the analysis of “clusters,” i.e. groups of countries that share good connectivity between them. Both studies —the one conducted in LAC and the one conducted in Africa— revealed a total of four clusters or sub-regions with good internal connectivity. For example, Argentina, Brazil, Paraguay, and Uruguay form one of the LAC clusters. These regions can be useful in deciding where to locate a certain service or content: a service located in Brazil could serve the customers located in the countries mentioned above. In terms of clusters, the biggest difference between LAC and Africa is that LAC clusters have lower latency values; latency among clusters is also lower.
The results of this study are the first step in a full connectivity analysis of the two regions. Future steps involve crossing latency data obtained by the Simón Project with other data sources (e.g. routing information) to determine which factors improve or affect connectivity.