[time-nuts] Timing performance of servers

David J Taylor david-taylor at blueyonder.co.uk
Sat Oct 27 11:05:17 EDT 2012

Lovely! I'm impressed.

What's the reasons for the offsets? Can't your tool handle negative values?

It would be good to have min, max, max-min, avg, std.dev values without
offsets to help illustrate worst-case behaviour as well as average
performance and noise "energy". The more advanced plotter would show
MADEV, TDEV and MTIE plots. Ah well.

Would it be possible to set up so you could measure deviation on SNTP
and undisciplined machines?

PS. Have my summerhouse not to far away from the town Ystad.


I'm glad the information was helpful, Magnus.

Yes, the tool I use - MRTG - can't handle negative values, and I haven't yet 
had the need or enthusiasm to convert to RRDtool which likely can.  I do 
also collect the normal NTP statistics on some PCs and wrote an NTP Plotter 
program to analyse those here:
Move your mouse over the list of plots below to see the different plots.  It 
doesn't include the more advanced statistics which you mention.

The plots in MRTG are from a Perl script which uses ntpq and interprets the 
output into the two numbers needed for an MRTG plot.  Another tool I offer 
is my NTP Monitor program:
This tries to use NTP or time protocol to determine the time on a remote PC, 
and plots it against a reference such as a local PC.  It's a coarser tool 
than using the NTP statistics, and was designed when my own time keeping was 
much worse, and I had no local stratum-1 servers.  You could possibly do 
something simialr in Perl and use that to feed MRTG.

PS. We know Sweden and Norway from holiday visits, and Ystad from the 
excellent Wallander TV series (in Swedish)!  A Northern Lights Norway trip 
is written up here:

SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor at blueyonder.co.uk 

More information about the time-nuts mailing list