I am wondering if an adversary would be able to derive useful information due to the fact that the consensus weights are changed abrupt ? (screen shot attached)
I also think that, once a server is running for weeks (or month(s) ?) it could be usefull if the bandwidth authorithy system is able to "lock" a good speed based on the server history. Instead of "rising, falling, rising, falling, rising, falling" in unstable loop, because if it's day or night, too highly loaded and, few hours after, underloaded, every every and every day.
It would unlock this value and re-evaluate it only if current test shows that this is really needed (suddenly excessive congestion, or too many consecutive tests that confirms that the server is more overloaded, or more underloaded that before). For example. But may be the Tor Project team already have a lot of work on other subjects ;)
----- Original mail ----- From: "Toralf Förster" toralf.foerster@gmx.de To: tor-relays@lists.torproject.org Sent: Vendredi 23 Janvier 2015 14:00:13 Object: [tor-relays] could smoothing/smearing a consensus weights change makes sense
I am wondering if an adversary would be able to derive useful information due to the fact that the consensus weights are changed abrupt ? (screen shot attached)
tor-relays@lists.torproject.org