On Jun 16, 2012, at 22:30, Ask Bjørn Hansen wrote: > The lines for the offset data were hiding useful information (at best) or > just making up information (at worst). The (green) dots show the actual data > points.
I wanted to show a couple examples of how I believe the scatter plots show more information: This is a NIST server; it's not in the pool DNS ever, I just added it to the system to monitor it, on the scatter plot chart you can clearly see what looks like two load balanced servers, right now one has a 1ms offset generally and the other one at 1.5ms. On the old line graphs it'd just be up and down noise. http://www.pool.ntp.org/scores/192.43.244.18 This is a Soekris Net4501 (very small CPU) box, keeps time really well (ntpd says it's 0.1ms off generally), but I think easily overloaded which makes the timing a little off: http://www.pool.ntp.org/scores/207.171.7.151 This is a Soekris Net4801, I wonder if it's placed somewhere so it gets more affected by temperature: http://www.pool.ntp.org/scores/207.171.7.152 -- the old line graphs would just show this as jagged up and down. This is a stratum 1 / GPS clock on a slow connection. Notice how there's no real pattern, just a "band" of offsets around 1-3ms (it's the DSL connection I mentioned some days ago). With line graphs it'd be hard to see if there is a pattern. Here it's clear there isn't. http://www.pool.ntp.org/scores/64.81.84.140 Ask -- Ask Bjørn Hansen, http://askask.com/ _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
