> 2. Why did the disruption to the Skype network begin on on Thursday > if the Microsoft patches starting updating customers' computers on > Tuesday?
That's easy, folks didn't get the reboots done until then. The patches come out. Machines begin downloading them. Users get a notification about the available upgrades. Users have work to do and Cancel performing said upgrades. Machines eventually get rebooted. Meanwhile the press made a lot of noise about security issues and the value of upgrades so the users actually get around to doing them. > 3. Why didn't the Skype "self-healing" behavior work? Pfft, is this where "self-healing" is proved more as a marketing buzzword than fact? > Was it due to > the unusual scale/scope? Was it due to reliance on centralized > resources that were themselves overloaded at that time? Is the same > mechanism used for failing-over live phone calls as is used for "self- > healing" server access? > > There are probably other questions we should be asking. Indeed, even with high speed networks, p2p and fast computers there are still very necessary efforts to be taken in planning and scaling systems. At least it's not like the old days where you couldn't reboot all the servers at once because the power load would cause a brownout. But it's certainly similar... -Bill Kearney _______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
