Jon, On Aug 25, 2025, at 8:36 AM, Jon Lewis via NANOG <nanog@lists.nanog.org> wrote:
On Sat, 23 Aug 2025, David Conrad via NANOG wrote:
The problem is the assumed binding of <IP address, geographic address> in the “northern hemisphere” or wherever. This has never been guaranteed, has always been questionable, and, historically, was actively discouraged, at least by the RIRs (“the Internet does not use geopolitical boundaries for address allocation”, handwaving away the RIR geographical monopolies).
Huh? It wasn't that many years ago, ARIN considered "out of region" utilized IP space to not qualify as "utilized" for purposes of qualifying for additional allocations by showing your existing allocations were sufficiently utilized.
RIRs behaved differently (adding to the confusion). Historically, the definition of geographic location for address blocks in the context of RIRs frequently devolved to the location of an organization’s headquarters, not where addresses were actually used. Worse, at least from the perspective of the routing system, in the very earliest days, multi-nationals were supposed to go to the RIR where their headquarters were, even if the address space was to be used outside of that region (e.g., from personal experience, Shell Oil Company in Indonesia were supposed to get address space from RIPE-NCC because Shell had their global headquarters in the Netherlands). Any documentation of location of use beyond “HQ” was dependent on the organization being truthful and conscientious, which in many (most?) cases, translated into a mad rush to fill in assignment information (honestly or not) just before applying for additional address space. I’d imagine ARIN’s policy led to a certain level of “creativity” and additional costs, both for the requesters who came up with their reassignments (or, in the best case, developed tools to keep that information up to date) as well as ARIN staff that would be tasked with verifying the sub-assignments.
The crux is that, when it doesn’t, the mechanisms to fix the binding, such as they are, sucks
This varies quite a bit from one IP Geo provider to the next. Some are pretty good (have web pages where you can do test queries against their data, will accept your geofeed data if you tell them where to get it, etc.). Others (like Digital Element) seem to be entirely opaque and obtuse.
Needing to figure out who to contact and then navigating the bespoke mechanisms in the hope that they will work, without any sort of mechanism to appeal if they don’t, in order to fix information that ISP’s customers are operationally dependent upon in a timely fashion sort of fits into my definition of “sucks”. I suppose this will get fixed after ISPs who lose customers sue the geo location providers for causing loss of business, so as usual, the lawyers always win. Regards, -drc