On Sun, Mar 23, 2014 at 11:50 PM, Hal Murray <[email protected]>wrote:
> > [email protected] said: > > The assumption NTP makes is that you can judge the quality of a server > by > > the variance (of "jitter") in the time it reports. > > I think it's more complicated than that. > > I think it also includes the non-jitter part of the round trip time. Yes. NTP calls it "root distance" which includes a jitter and delay component. I think (?) they project jitter and delay on a plane and compute the distance to a centroid and call that "root distance". But it does have both components as you say. I just re-read the docs. The point is that NTP's assumption that one way equals 1/2 the round trip if wrong causes some inaccuracy. I think you are right about a fixed offset. But if there are other clocks that are better they will cluster and the one with the asymmetry will be removed from the set that is used. The offset will cause it to be an out layer NTP is pretty good at detecting bad clocks if it has enough clocks to compare. I always like to use at least five. > NTP > assumes the path is symmetric. Any constant asymmetry will turn into an > apparent fixed offset. I think ntpd is smart enough to include that in > it's > calculations of the clock quality, but I don't understand the details. > > > -- > These are my opinions. I hate spam. > > > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > -- Chris Albertson Redondo Beach, California _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
