MUMBAI — As the internet infrastructure of the Asia-Pacific region continues to evolve amidst rapid technological shifts, network operators, researchers, and infrastructure experts gathered in Mumbai, India, for the APNIC 62 conference. Held from September 4 to 10, 2026, the week-long event served as a central hub for discussing operational experiences, emerging technologies, and the persistent operational challenges facing modern internet management.
Among the conference’s prominent tracks was the Cooperation SIG session, chaired by Joy Chan alongside panellists Sid Mathur, Shailesh Gupta, and Dhruv Dhody. The session, titled “Behind the pin: The reality of IP geolocation,” took a deep dive into the multifaceted complexities of determining the geographic location of Internet Protocol addresses. Panellists and presenters examined the limitations of traditional registry data, the emerging role of geofeeds, the perspectives of commercial geolocation providers, and the ongoing, industry-wide search for more accurate and privacy-preserving approaches to network mapping.
The Evolution and Limitations of IP Geolocation
Geolocation has a long history on the internet. Regional Internet Registry (RIR) Whois and Registration Data Access Protocol (RDAP) services, alongside joint RIR statistics files, have traditionally provided authoritative information regarding the entities holding specific IP address resources. However, experts at the session noted that this data is frequently misunderstood by the public and downstream users as indicating where IP resources are physically used, when it actually records where they are legally registered.
While Whois and RDAP have long supported specific registrations for IPv4 and IPv6 resources—allowing holders to assign different ISO 3166 two-letter economy codes to distinct prefixes or address ranges within their holdings—true geographic determination relies on entirely different mechanisms. These include formal frameworks such as the RFC 8805 standard for publishing geographic data and the RFC 9632 standard, which provides a reliable method for discovering that published information.
AIORI IP Geolocation and Geofeed Analytics
Opening the technical presentations, Anand Raje shared insights from the Advanced Internet Operations Research in India (AIORI) project, detailing ongoing work to measure, validate, and analyse IP geolocation data. The project is actively developing an advanced platform combining measurement data with published location details to aid geofeed searches and deep analytics.
A central thesis of Raje’s presentation was that IP geolocation must always be treated as evidence rather than absolute certainty. While an IP address can frequently be linked to a specific economy, region, or city, it does not necessarily identify an individual person, a physical device, or an exact street-level location. The practical accuracy of any mapping effort ultimately depends heavily on the chosen data source and the confidence level that can be placed within it.
Raje pointed out that modern geolocation data underpins an enormous range of critical decisions, stretching from content delivery optimization and traffic engineering to fraud detection, legal compliance, international sanctions screening, and public safety. As the real-world impact of these automated decisions continues to grow, the operational and financial costs of geolocation errors rise correspondingly.
To tackle these challenges, the presentation argued that no single technique can reliably determine location on its own. Effective systems must instead combine multiple signals, including registry and routing data, active measurements, machine learning algorithms, operator-published geofeeds, and client-provided information. To support this ongoing research, AIORI has deployed approximately 500 geolocated measurement anchors across India, alongside an anycast testbed spanning more than ten distinct locations, providing independent evidence to assess and benchmark geolocation accuracy.
Despite these efforts, Raje highlighted a notable disparity in geofeed adoption. Deployment rates within the APNIC region remain significantly lower than European levels, with the majority of published geofeed prefixes currently originating outside the Asia-Pacific area. Furthermore, analysis of existing geofeed data revealed wide variations in geographic precision, with many records providing only city-level information while relatively few incorporate street-level detail—reflecting both the inherent technical limits of network mapping and conscious efforts to safeguard user privacy.
IP Geolocation IAB Workshop Report
Providing a macro-level perspective from the Internet Architecture Board (IAB), Dhruv Dhody presented findings from an IAB workshop held in late 2025. The workshop offered a comprehensive overview of the Internet Engineering Task Force (IETF) viewpoint on the geolocation problem space, detailing current mechanisms and highlighting lingering architectural gaps.
As Dhody emphasized during his presentation, IP addresses were never originally designed to carry geographic meaning. They serve primarily as geography-neutral identifiers intended to provide unique addressing within Internet Protocol networks. These identifiers are carried inside data packets and routed globally using core protocols such as Border Gateway Protocol (BGP), Open Shortest Path First (OSPF), and Intermediate System to Intermediate System (IS-IS).
Over time, however, these technical identifiers have become inextricably linked to real-world questions about physical location. Governments, commercial enterprises, academic researchers, and everyday users increasingly demand to know precisely where internet traffic originates and where it is destined in a physical, geographic sense.
Consequently, existing approaches are coming under severe strain as internet connectivity grows increasingly dynamic. Low Earth Orbit (LEO) satellite constellations, multinational mobile network operators, and highly distributed service architectures all make precise location mapping considerably more difficult. In many modern scenarios, users seamlessly move between distinct access networks while their application sessions remain continuously active, amplifying the pressure to improve underlying geolocation technologies while simultaneously protecting user privacy.
Self-Published Geofeeds and the Role of RIRs
Sid Mathur addressed the Cooperation SIG panel and later Tech Session 2 (AI), focusing on the geofeed publication and discovery mechanisms defined in RFC 8805 and RFC 9632. He also demonstrated how artificial intelligence tools can be leveraged to improve data quality by automatically identifying inconsistencies and assisting network operators in proofreading published geofeed data.
Mathur argued that RIRs occupy a unique and advantageous position within the global geolocation ecosystem because they already operate the authoritative registration systems used to identify resource holders. Rather than inventing entirely new bureaucratic processes, he suggested that the existing registry framework could serve as a trusted foundation for geolocation publication. By building on workflows operators already understand, this approach creates a verifiable chain of trust through RFC 9632 and represents a high-leverage opportunity to drive broader awareness and adoption.
Highlighting regional imbalances, Mathur pointed to data showing that 75.5% of successful geofeed publishers were located within the RIPE NCC region, compared to 21.6% in the ARIN region and a mere 3.0% in the APNIC region. To bridge this divide, he called for consistent support from all RIRs, pointing to RFC 9877—co-authored by APNIC and ARIN and published in October 2025—as an existing mechanism designed to simplify deployment. His primary recommendation was straightforward: integrate geofeed publication as a simple, one-click function directly inside member portals and registry dashboards, noting that an informal APNIC community survey identified uncertainty over registry assistance as a primary barrier to adoption.
The ISP and Commercial Data Provider Perspectives
Shailesh Gupta presented an Internet Service Provider (ISP) perspective, outlining the complex landscape shaped by major geolocation providers that combine RIR data and self-published geofeeds with proprietary datasets and independent methodologies. Because organizations seeking to correct geolocation errors often face multiple fragmented reporting channels, the process can become unnecessarily convoluted.
Gupta stressed that geofeeds should always be understood as a best-effort mechanism whose accuracy is governed by diverse factors, including shifting update cycles, widespread VPN usage, and evolving network deployment strategies. Often, geolocation information is intentionally approximate to preserve user privacy, identifying only a state or city rather than precise coordinates. Furthermore, accuracy generally scales downward from reasonably reliable ISO country codes to less reliable state, regional, city, and street-level data.
Adding to the complexity, Gupta highlighted the challenge of place-name ambiguity during the AI session. Names like Frankfurt or Kendal can refer to multiple distinct locations globally, creating data quality issues that cannot be resolved merely by updating a geofeed if the third-party provider misinterprets place names and maps them incorrectly to network infrastructure.
Luna, product owner for GeoIP services at MaxMind, offered the perspective of a third-party data integrator. She explained that self-published records represent only one category of information utilized by commercial firms, which must synthesize BGP relationships, network usage patterns, browser-derived signals, and other operational datasets to determine location. This reality raises persistent questions about what truly constitutes ground truth for geolocation, as different stakeholders maintain distinct expectations. Luna acknowledged notable regional discrepancies in data volume and quality between Western markets and parts of the Asia-Pacific, Africa, and Latin America, while outlining practical steps that regional ISPs can take to collaborate with providers and improve overall accuracy.
The session concluded with a shared consensus that IP geolocation is no longer a simple mapping exercise. As network infrastructure grows increasingly distributed across cloud environments, mobile networks, and satellite systems, maintaining accuracy will require transparent collaboration, wider geofeed adoption, and robust validation mechanisms that balance operational needs with user privacy.
Leave a Reply