detects geolocation hints embedded inside hostnames and builds
regular expressions (regexes) to extract them.
Figure 1 illustrates the approach by example, where there
are six hostnames for three routers with geolocation hints
embedded, which are feasible given delay constraints from
vantage points with a known location.
Conceptually, we conclude the regex accurately represents
the naming convention for the suffix if the regex usually
extracts an apparent geolocation hint from the hostname, and the regex
behaves this way for numerous routers with hostnames in the suffix.
We are at that stage where we need feedback from operators
to validate/improve our algorithm. We include below a set of
inferred geolocation conventions for routers in approximately
one thousand domains.
If you would be willing to donate a few minutes of your time
to validate inferences (or let me know of errors) for any
domains you are intimately familiar with, please contact
mjl@caida.org.
Figure 1: Naming Convention for zayo.com.