[OpenIPMap] Changes to OpenIPMap last week

• Next message: [OpenIPMap] Geolocating remote peerings at IXPs
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Over the easter-weekend I added some functionality to OpenIPMap to
improve the geoloc-suggestions the system offers, and I put this live
last week, so it's about time to say something about it.

The OpenIPMap traceroute visualiser (at
https://marmot.ripe.net/openipmap/tracemap ) has 2 types of geoloc for
IPs:
- - crowdsourced information (ie. the info you put in)
- - geolocation guesses the backend-system makes.

Previously system guesses would be shown with red background and show
up as red dots on the traceroute visualisation, and the crowdsourced
info in green.

What I added is that the system guesses are now restricted to things
that are possible RTT-wise.

Earlier the guessing was mainly based on trying to find full city
names, IATA or CLLI codes in hostnames or trying to figure out if
parts of the hostname are abbreviations of city names. If multiple
cities were found, the city with the largest population was offered as
a guess (using population was inspired by actual research:
http://nowak.ece.wisc.edu/geoloc1.pdf ).

Now if multiple cities are found, only the cities that make sense from
the RTTs in the current traceroute are returned (ie. a hop 10 ms from
an Amsterdam RIPE Atlas probe is not going to be in Sydney, but it is
going to be within ~1000km from Amsterdam).

Also, if the system finds multiple possible cities, it is going to
return the city with the most geolocated entries in the OpenIPMap
database, instead of the city with the largest population. Now that
OpenIPMap has a significant body of crowdsourced geoloc this starts to
make more sense.

Users interacting with the system might have also noticed that the
system guesses are not coloured red anymore, but orange instead. The
text is prefixed with '??' so it hopefully is more clear to the user
that these are guesses.

Martin Levy also offered some patches for better markup of traceroutes
in the user interface.

cheers,
Emile

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=b/3i
-----END PGP SIGNATURE-----

• Next message: [OpenIPMap] Geolocating remote peerings at IXPs
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]