Geo location (or database lookup?) problem

[iamback]

Just found this (project_id=42910:

Location:  Flevoland, Almere, Spain
Hostname:  213.194.44.30

Now, according to NERD-ZZ 213.194.44.30 is indeed located in Spain - see:

• http://www.moensted.dk/spam/?addr=21...&Submit=Submit
• http://countries.nerd.dk/isolist.txt

But then where did that "Flevoland, Almere" come from? Almere is a city in the province of Flevoland in the Netherlands...

Also, looking at 'Country/State/City/ISP' (City) I see 5 hits from "Flevoland, Almere, Netherlands - from that same IP address which is actually in Spain!

In the same list (either City or Country), a visit from Korea (211.215.9.28 - that really is Korea) doesn't appear - possibly also misclassified under another country although it's listed correctly under recent page loads.

It seems there are some inconsistencies between what countries/cities visits are assigned to between different 'views' of the same data.

[webado]

It appears that geo-location databases are updated regularly with the information supplied by the ISP's themselves. But IP allotment and corrections are something that continually change, whereas the geo-location databases may not be published with as much frequency.

If you check an IP address today you may not get exactly the same geo information that you would have had yesterday or that you may get tomorrow in some cases.

[iamback]

Originally Posted by Hans

According to www.ripe.net/whois this IP is located in The Netherlands.

inetnum:      213.194.0.0 - 213.194.63.255
(...)
netname:      ES-CONCERT-19990920
(...)
country:      ES # IT NL

Note that both the netname and the country actually suggest it's Spain (ES), nevermind who the owner is. Also note that the record was changed twice since I made my post... And Senderbase (which isn't always up to date either) gives a different owner entirely.

All of which doesn't explain why two different views at the same moment here on StatCounter disagree as to country, and why it could combine a placename in the Netherlands with Spain as a country. Or why one view locates a hit in Korea (other sources agree) and another view at the same moment doesn't.

There definitely is some bug here, even if WHOIS records change (which I know they do). I'm sure StatCounter doesn't get its country information dynamically, but from a local database. So why should two lookups of that database for the same visit come up with different countries? In (at least) two cases? On the same day, even?

Originally Posted by chrisooc

It appears that geo-location databases are updated regularly with the information supplied by the ISP's themselves. But IP allotment and corrections are something that continually change, whereas the geo-location databases may not be published with as much frequency.

Sure. But that doesn't explain the inconsistency between the StatCounter results. While Geo location data may be uncertain, and change from one day to another, at one particular moment there's only one country for one IP address in teh database. If StatCounter comes up with two different countries for the same visit (and in at least two cases), then there's a bug at StatCounter's end.

[webado]

I don't know what to say. From my observations there is a difference between the base location associated with the ISP and the location associated with the actual Ip address logged. The latetr is the locaiton of the user, whereas the former is the location of the host.

So that particular ISP may well be headquartered in the Netherlands, however they may serve a wider area of Europe (most likely) and their particular user who got assigned that IP address for that particular visit was in Spain.


The country database which I use also places the Ip address in Spain. I get no actual ISP name from this IP address, in any reverse lookup I've used.

[iamback]

Originally Posted by chrisooc

I don't know what to say. From my observations there is a difference between the base location associated with the ISP and the location associated with the actual Ip address logged. The latetr is the locaiton of the user, whereas the former is the location of the host.

Yes, there can be.

But even that doesn't explain why StatCounter comes up with different countries for the same visit in different views of the same data!

I don't really care whether it's Spain or the Netherlands (hard to decide in this case), or whether it's Korea or somewhere else (unlikely) - as long as it's reported consistently, which it isn't. And a combination of a place name and province in the Netherlands with Spain as country is totally off - it can't be both!

Only one conclusion possible: this is a bug at StatCounter.

[iamback]

Ping webmaster!