Koos van den Hout wrote: > On Sun 2010-01-17 at 20:42, Anssi Johansson wrote: > >> Hi guys, I thought I'd try a slightly different approach towards Turkey. >> I know there have been efforts to get Turkish Telecom / TTNet to reduce >> their traffic, but even if they start to cooperate, getting the traffic >> down to reasonable levels will probably take quite some time. > >> Actually, it looks like the clogged pipe is the connection between >> Turkey and Romania. All the traffic from Tellcom to everywhere else on >> the planet seems to be routed via Romania. Traffic within Turkey is fine. > > Thanks for your effort. I was wondering if specific software used by TT > customers has to do with the high amount of ntp traffic, and on second > thought I am also wondering whether the clogged pipes between Turkey and > the outside world (where most ntp pool servers are) cause part of the > problem. I could imagine ntpd (or ntpd-like implementations) doing a few > extra queries when the answers get dropped on the way back. > > Koos
It's in extremely poor taste for an NTP implementation to send packets more frequently when a server appears to be down, since the only thing more fun for a server than being overloaded so much it's dropping packets is its traffic suddenly spiking at the same time! Of course, there are many poor NTP implementations, so it does happen. But the good ones (including the reference implementation) are better-behaved. -- Matt Nordhoff _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
