Okay, I have more detailed stats to report. This comes from exactly 24 hours of data; I collected a bit mroe and manually trimmed it to cut off right at 24 hours.
Perhaps the most interesting plot is the one of a histogram of packet counts grouped into ten-second buckets (grouped by the minutes and the tens digit of the seconds). In Fourier-transform terms, this plot shows substantial power at one minute, five minutes, 30 minutes, and one hour, with the 5-minute and 1-hour the most pronounced. (Because of the way it's grouped, of course, only divisors of one hour will show on this plot.) The plot is at ftp.rodents-montreal.org:/mouse/misc/ntp.10sec.gif for anyone who's interested. The data this was based on was collected by tcpdumping for "port 123" on my border. There may be some spurious data, if someone is firing NTP queries at something other than my pool host. A quick tcpdump for "port 123 and not host 216.46.5.9" for a minute or so makes me think that this traffic is so low as to be totally ignorable for the purposes at hand. If anyone would like to massage the data differently, I can provide the raw data file - but you'd probably be better off collecting your own data.... /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML [email protected] / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
