In this episode of PING, APNIC Chief Scientist Geoff Huston explores a recent issue with how IP addresses are geolocated in the APNIC Labs measurement system. To compile per-economy and regional statistics, and to analyse how experiments are distributed across market segments, APNIC Labs relies on free geolocation databases from MaxMind and IPinfo. These databases are built using various data sources, including Border Gateway Protocol (BGP) routing information, resource distribution reports from Regional Internet Registries (RIRs), whois and RDAP records, and self-published RFC 8805-format statements from ISPs.
At best, IP geolocation is only an approximation. As IP address mobility increases globally, it’s becoming more difficult to confidently determine the precise location of an IP address observed in Internet traffic. This challenge is further complicated by the widespread use of Virtual Private Networks (VPNs) and privacy-enhancing technologies like Apple Private Relay and Cloudflare Warp. While these systems aim to reflect the general geographic distribution of their users, they do so in a coarse, privacy-preserving manner — as Geoff notes, they make a best-effort attempt within those constraints.
Geoff was recently contacted by Ben Roberts of Digital Economy Kenya, a new AFRINIC board member and long-time industry analyst and technical advisor, who had identified anomalies in Internet statistics reported from Yemen. These figures didn’t align with the known realities of the region’s Internet economy. This prompted Geoff to take a closer look at how Starlink is influencing the distribution of Internet traffic. As a result, he’s made adjustments to APNIC Labs’ geolocation measurement practices. These changes will gradually become visible in the Labs statistics as the updated smoothing algorithms take effect.
Low Earth Orbit (LEO) satellite delivery of the Internet has had rapid and sometimes unexpected impacts on global Internet visibility. Due to orbital mechanics, nearly every part of the Earth’s surface is now Internet-enabled, although access often comes at a cost that exceeds what many in local economies can afford. This shift is fundamentally changing how we observe and interpret Internet usage, and it helps explain some of the discrepancies that have been emerging in the APNIC Labs data model.
Read more about Geolocation and Starlink on the APNIC Blog and on the web:
- Geolocation and Starlink (Geoff Huston, APNIC Blog September 2025)
- RFC 8805 ‘A Format for Self-Published IP Geolocation Feeds‘ (IETF RFC website)
- The NRO RIR Statistics on delegations with geographic tagging of the delegated entity (NRO website)
- MaxMind GeoIP resources (MaxMind website)
- IPinfo.io (IPinfo website)
- Labs statistics portal (APNIC Labs website)
Subscribe and share your story
You can stream and subscribe to PING via the following channels:
If you’re interested in sharing your insights or research, please get in touch — we’re always looking for great stories from the community. Please let us know what you think of the podcast and the APNIC Blog so we can keep improving.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.