[ntp:questions] NTP makes a time jump
stenn at ntp.org
Mon Jul 8 19:26:08 UTC 2013
John Hasler writes:
> I mean so that none ever have to slew at the 500PPM max rate.
This one is interesting - in a group of machines each machine may have a
different natural drift rate. If the clock adjustment algorithm doesn't
have a way to handle adjusting the functional equivalent of 'tick' then
you can get into situations where, at the worst case, some clocks are
running fast (say, by 250ppm) and some are running slow (say, by 250ppm)
and at that point there is at most 250ppm of "room" for adjustments.
Harlan Stenn <stenn at ntp.org>
http://networktimefoundation.org - be a member!
More information about the questions