[ntp:questions] Jitter versus polling interval

David J Taylor david-taylor at blueyonder.neither-this-bit.nor-this.co.uk
Sat Mar 7 08:49:04 UTC 2009


David Woolley wrote:
> Danny Mayer wrote:
>
>> Increased polling frequency increases jitter.
>
> I think jitter was being used loosely here, not with the specific
> sense of high frequency variations.

Yes, my original question referred to looking at a graph of offset plotted 
over a day (sampling at 5-minute intervals) and trying to judge the amount 
of short-term oscillation as opposed to the longer-term drifts (e.g. those 
caused by temperature variations).  I was not meaning the "jitter" figures 
reported by NTP.  I was actually thinking about my Windows Vista system:

  http://www.satsignal.eu/mrtg/gemini_ntp.html

where the average offset (on a scale from -100 to + 100ms) is zero, but 
there is noticeable jitter.  I was playing with maxpoll, and the jitter 
seems worse with increased polling interval.

The performance with Vista is now substantially improved since Dave Hart's 
version became available, and I am seeing good performance making a 
Windows system into a stratum 1 server with a GPS/PPS source.  I may even 
be able to switch off my old FreeBSD system (which have given excellent 
service).

Cheers,
David 




More information about the questions mailing list