How do you make a geolocation database?

Question

I am not asking what geolocation service to use or how you use them.

I am asking, how the do these companies know so well where every IP address is?? Is there some breach of privacy being violated?

I looked at the wikipedia page, and all they had to say was using the WHOIS service, which obviously doesn't work at all: my IP is owned by a company listed in another state.

As a sidenote, when I was making this question, Stack Overflow popped up this banner:

The question you're asking appears subjective and is likely to be closed.

Answer 1

It has a lot to do where the ISP's are logically located and that ARIN knows where networks are assigned.

They can also determine your location based on routers.

run this in a command/terminal window : tracert google.com

I'm sure you can see some location based info in your tracert.

Answer 2

Those companies pay for the data.

There are many ways to get this data (not all illegal), one simple one is, for example, providing free services that encourage you to provide some information about your actual location like for example DslReports. Once they know one IP and the ISP is easy to correlate other IPs from the same area.

As you can see here one company recommends the other so you can see the connection.

Answer 3

Many such databases appear to be extracting it from the 'whois' databases held by the Regional Internet Registries (RIPE, ARIN, etc).

These are not the same at the domain name 'whois' lookups, these relate specifically to IP addresses.

Such data extraction is an illegal breach of their database copyrights and strictly against their T&Cs.

See http://stackoverflow.com/questions/274308/how-does-geographic-lookup-by-ip-work for more details.

Answer 4

Let me answer with an analogy each car has a unique number that identifies it from it's manufacturer, the company has a list of all the cars that where send to each mayor distributor on each part of the world, each one of those distributors has several dealers to which they assign a set of cars to sell, and each one of those dealers sells the cars to end customers. So in theory if the manufacturer wants to know where is the world is a car he doesn't has to ask because he know in which country it landed.

Translating that to IPs every company that sells public IP address has a record of who owns it, and they are normally give them away in bulks of 1000s to ISP (phone numbers used to be like this). For example I can tell you if an IP is from my country just by looking at the first 2 groups. On the other hand hosting providers and data centers work the same way and they almost always know where is the machine physically, and last but not least doing a trace will jump hops to the closest (theoretically, as you can force the traces to be what you want) IP to the box which means you can guess the location if you have the one of the hop before it.