Skype's vagueness generates suspicion, but to deny an actual D-o-S attack would seem a pretty dumb and risky thing to do.
Some wild uninformed speculation about "why now and not previously?": it may have been an accumulation of "patch Tuesday" restarts over several months that exercised their bug. For example, what if a potential superpeer reports its average uptime for the last N (N>1) sessions? And this latest widespread restart, combined with one or more previous ones, was able to push a critical mass of machines' values under some system-critical (even if arbitrary) uptime threshold? Perhaps, there's a rule that says "always make sure you have one superpeer whose last 3 uptimes average over 30 days". And, in the early days of Skype, they always ran enough of their own nodes with this quality to heal usual disruptions. But with time and growth, Skype became hooked on easy p2p credit^H^H^H^H^H^H leverage to anchor their network. And then when an insufficiently-modeled "black swan" uptime-liquidity-crunch occurred, boom. Perhaps their next blog post will explain how this level of outage was a "25 sigma" event expected "once every 10,000 years". - Gordon _______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
