> From the paper by Murdoch and Zieliński [3]: > [3] http://www.freehaven.net/anonbib/#murdoch-pet2007 > “We suggest that existing models, based on Autonomous System (AS) > diversity, do not properly take account of the fact that while, at > the AS level abstraction, a path may have good administrative domain > diversity, physically it could repeatedly pass through the same > Internet eXchange (IX).”
It would be interesting to include a traceroute service in each node such that path building might take into consideration such IX/AS repetitions over the proposed full path. Note however that MPLS routing used to engineer traffic schemes can hide the actual path from traceroute (though mitigated by its usual use only within one AS). And other than of geolocation, it's quite hard to align IX nodes with traceroute data. Last, if the overall src and dst fall within purview of your tapping GPA of choice the usual timing/counting shortcuts apply. It would be rather interesting to see if transiting the Tier-1's could be avoided altogether by pathing circuits through the side peerings that the lower tiers establish between themselves, as discovered by said traceroute mechanism. Though that might work within a continent, don't ever think it will get you between them safely... cable landings are far too ripe, as is any model where a long haul fiber owner leases out strands to multiple Tier-n's... that shared bundle terminates somewhere, and often at many places along the way. See NANOG for further Internet background. [Somewhere in all of this there is something to be said for peer owned p2p meshnets and co-op's...] _______________________________________________ tor-talk mailing list [email protected] https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk
