LACNIC promotes different types of Internet measurements within the region and in relation to connectivity throughout Latin America and the Caribbean.
These measurements help us to understand the state of network connectivity among countries in our area of influence and allow operators to use the information to improve the performance of their networks and interoperability.
In 2020, we conducted a first study on latencies, both within and between countries, which provided us with an approach to the state of connectivity in the LAC region.
This initiative generated inquiries and awakened the interest of the community, motivating us to promote further research based on available routing tables in the region: BGP interconnection in the Latin American and Caribbean region and BGP interconnection at the local level in the Latin American and Caribbean region. These new studies allowed for a broader and more comprehensive view of routing-level connectivity in each country and at the regional level.
Connectivity Measurement Study 2023
The information gathered and community feedback led us to undertake a new project to actively measure traffic behavior using traceroutes. This study [in Spanish] serves two purposes: it is longitudinal because it allows for comparisons with previous studies conducted using a similar methodology in past years, and it is also cross-sectional as it provides a descriptive overview of a large number of networks in the region at a given moment. Traceroutes not only measure latency from the source to the destination (end-to-end) but also reveal the intermediate networks through which information travels. And with the results, it is possible to analyze which ASNs (Autonomous System Numbers) have greater centrality in each country, whether the paths are of reasonable lengths, whether the queries stay within each country or involve interconnections outside, and whether the presence of local IXPs (Internet Exchange Points) is observed in that country, among other data.
The work sought various sources of information in relation to the other studies. In the previous measurement, we limited ourselves to latency times between countries; now, we extend it to the network or ASN level.
We incorporated information from RIS (RIPE Routing Information Service), which allowed us to translate the IP addresses found in the traceroute into ASNs. This allowed us to immediately identify which ASN traceroutes were passing through when observing suboptimally routed traffic on the network.
The report includes relevant information about the packets that pass through the IXPs and those that do not, thanks to the PeeringDB data source. This data enables us to analyze the routes passing through IXPs compared to those that do NOT. As a result, we were able to measure the influence of IXPs within each country.
We identified countries with very favorable results where over 40% of the launched traceroutes pass through the local IXP. This is the case for Bolivia, Panama, Ecuador, Argentina, Costa Rica, and Colombia. Additionally, Argentina, Ecuador, Chile, Bolivia, Haiti, and Nicaragua were the countries that had more internal routing. However, there is a long list of countries that have quite good routing.
The results also show the ASNs with greater centrality in each country, whether the paths have reasonable lengths and if the traceroutes remain within each country or involve external interconnections, among other data.
The study also offers a quadrant comparing two dimensions: the penalty paid by outgoing traceroutes in terms of latency, vs. the number of outgoing traceroutes. A positive observation is that there are no countries that exhibit high latencies and high number of outgoing traceroutes at the same time. However, there are cases that display some of these properties. There are also the cases that exhibit none (where routing is relatively optimal). These are the cases of Chile, Uruguay, Paraguay, Mexico, Brazil and Argentina.
The study focuses on providing a broad overview of the state of connectivity in the countries and networks within the region and does not emphasize specific research investigations. In this sense, it is of utmost importance that local readers interpret the results of the study and contrast it with their local knowledge.
Request for Feedback from Operators
We invite operators to take a closer look at the study, look at the quadrant of outgoing and incoming routes and connectivity to IXPs, and note the latency penalty incurred when traffic leaves their country. You may contact us at the following e-mail address: imasd@lacnic.net
If you want to see the presentation made on this topic at LACNIC 40- LACNOG 2023, please click here.
