Geo-IP going the way of the dodo?
Managing 10,000+ endpoints on broadband over the past 5 years, you start to notice some trends. One of those trends is how incompatible GeoIP is with the adoption of mobile networks.
We have over 4,000 mobile broadband devices across the US and there are some very interesting things that have come up because of that:
- Most IP addresses for mobile phones/cards come from a pool of IP’s assigned to a POP for that carrier. Those POPs tend to be aggregated in the mid-west/central areas, for instance AT&T has centers based out of Texas, Illinois and Utah. Those POPs are responsible for issuing IP address for -all- mobile devices on the network, thus the IPs are registered for that locality.
- With IP addresses coming from centralized locations on the network, GeoIP enabled applications basically fail horribly at detecting where your load is coming from. That includes CDN services like Akamai & Limelight.
- Applications that try and be smart like Google Maps, home brew or web based apps will eventually fall down for not properly identifying the users locality.
There is only one solution, really, and that is to integrate another layer of LBS (location based services) that depends on triangulation of the signal or Lat/Long coordinates from the devices GPS. Some mobile providers are starting to implement LBS for their mobile devices but they are focused on the enterprise customers locating their devices instead of allowing the site/application to get that data.
Eventually it comes with a security trade-off which most customers will probably decline. The only lasting solution is for all of the mobile providers (in concert with landline & wireless providers) to agree on a single location specification which will identify users down to a specific X mile radius, zip code or Lat/Long of the closest city. This should give companies who care the ability to track their users and provide them with contextually relevant information/advertisements while keeping customers actual locations private.
I am very interested in how this will play out and what other software chicanery will show up trying to fix this natural problem of the globalization of wireless based connectivity. MiFi/Mobile Hotspots are just the beginning. With LTE/WiMax rubber hitting the road all over the place – wired internet access will end up being the bastion of always-on stationary computers/media devices and datacenters. Even my office is considering a wireless backup circuit for connectivity – go figure.
