> By Matthieu Latapy
The RTT (Round Trip Time) between a computer and another on the internet basically is the time needed for a piece of data to go from the first computer to the second, and come back. To this regard, it may therefore be seen as a rough measurement of a distance between two computers.
I picked random IP addresses (i.e. random 32 bit integers) and then measured the RTT from my computer to them. In many cases, I received no answers (as there was no answering computer at the random address), but I insisted, and I finally obtained 1 109 182 answers, and thus this many RTT measurements.
The plot above shows the observed RTT distribution, i.e. for each possible value of the RTT the number of times I observed it in my experiment.
Surprisingly enough, the plot exhibits clear peaks.
I guess this is a consequence of the fact that there are large numbers of computers at a similar distance from mine (for instance, french computers are very close, european computers are close, american computers are a little further, asian computers even further, and african computers are very far) (remember that the notion of distance here has not much to do with kilometers, but with internet speed). Conversly, there are some distances at which I will probably not find any computer (at the distance corresponding to the time needed to go to the middle of the Atlantic, for instance).