[ntp:questions] Re: NTP Synch >125msec Delays?

Brian Utterback Brian.Utterback at Sun.removeme.COM
Tue Nov 11 18:49:43 UTC 2003


Dairiki, Robert A (US SSA) wrote:
> The NTP.ORG FAQ says "time difference of less than 128 ms between server
> and client is required to maintain NTP synchronization".  The FAQ also
> indicates its not clear on the exact effect of >128 ms delays.
>  
> My question is what happens with >128ms delays, for examples, with
> network delays of 300-500 ms. 
>  
> Thanks.

You are confusing "difference" with "delay". The significance of the 128ms
is mostly one of definition, i.e if there is greater than a 128ms offset
between two clocks, they are not "in sync". This is not entirely arbitrary;
it is the vlaue used in the correctness proofs. If the offset is less than
128ms, and the drift is less than 500ppm, then NTP will be able to get the
clocks closly in sync. Otherwise, all bets are off, it might do it, or not.
So, in the case that the offset if greater than that, a step is performed
and all data is thrown away, making NTP "start over".

While the delay is important, it doesn't really matter how long it is, as
long as it is less than the current polling period and that the round-trip
is symmetric.

-- 
blu

Lesson from the blackout of 2003:
The power grid is THE most critical infrastructure, upon which all
others depend, and nobody really knows how it works.
--------------------------------------------------------------------------------
Brian Utterback - Solaris Sustaining (NFS/Naming) - Sun Microsystems Inc.,
Ph/VM: 781-442-1343, Em:brian.utterback-at-ess-you-enn-dot-kom




More information about the questions mailing list