CONICET | Buscador de Institutos y Recursos Humanos

In this paper we study the impact of the implementation of an IXP in the city of La Paz, Bolivia. The case has some desirable characteristics. First, Bolivia ranks at or near the bottom in Internet development (however measured) in Latin America, and thus presents a fertile environment in which the benefits of an IXP should be readily observed. Second, the IXP emerged from a new interconnection regime that requires that large access providers exchange traffic locally through a single interconnection point. This is critically different from other scenarios in which IXP membership is voluntary, and thus its impact is highly dependent on which networks choose to participate. Our impact evaluation strategy is twofold. First we analyze industry performance data collected by the Bolivian telecom authority. Second, we deploy active probes located in different network operators across Bolivia in order to obtain direct measurements about latency, the utilization of local routes, and other network performance metrics. These are combined with traffic indicators obtained from the IXP itself. Our preliminary results suggest that PIT Bolivia has made a positive impact on network performance in Bolivia. More specifically, local traffic passing through PIT Bolivia exhibits significantly lower latency and fewer hops than local traffic routed over international transit links. However, because traffic levels at PIT Bolivia are low and growing slowly, the economic benefit to participating ISP, and therefore to the Bolivian Internet ecosystem as a whole, is more modest than predicted. Further, given existing Internet usage patterns that heavily favor content access, the current absence of CDNs at PIT Bolivia considerably limits traffic growth, thus reducing the potential impact of the initiative.