Towards measuring content locality

Local access to Internet content is generally known to be cheaper and faster, but a lack of local content affects users' ability to access other resources, for example in cases where utility bills are paid online. However, despite this importance, we lack a comprehensive framework to perform such content locality analysis. In this work, we present a framework and provide preliminary evidence to support its effectiveness. To quantify the extent to which Internet traffic stays local in a country, we use the list of the Top 1000 most popular websites from Google's Chrome User Experience Report (CrUX) as a proxy to measure traffic locality. Our solution removes censored sites per country and determines whether the remaining websites are hosted natively on a server or on a Content Delivery Network (CDN) cache. We then find the location of the server or CDN cache using a mix of techniques, including geo-hints extraction, IP geolocation, and website location sourcing.