[ntp:questions] Re: Small fixed offset

Danny Mayer mayer at ntp.isc.org
Sat Oct 29 04:10:38 UTC 2005


ariel.burbaickij at gmail.com wrote:
> Neither, nor.
> 
> We have network in central part of USA. There some traffic
> is flowing. Monitoring equipment is located in Central Europe (there
> are plenty of reasons for placing it there). Signals from
> USA are backhauled via optical fibre cable to this
> monitoring equipment. Monitoring equpment represents
> NTP client which gets the time from NTP server also located
> in Central Europe. All signals from network in States are timestamped
> at their arrival at the interface of monitoring equipement. Now, as the
> timestamp should show the time as the signals were indeed on wire
> in States and not the time they were on wire in States + network
> latency time   the clock on the monitoring equipment must be
> slightly behind (150 ms) from the real time. So how we intend
> to do this is to feed the monitoring equipment with the time
> from NTP server with GPS unit locally attached with the the fudge value
> of -150 ms.  Is it workable approach?
> 

I think that what you really want is to make adjustments to the
timestamps to deal with the time it takes to send packets to and from
the US. I think what you really should do is to take the calculated
offset from ntpd and apply it to the logs rather than the other way
round. There are ways to get to the offset information so that they
estimated "corrected" time can be calculated for some definition of correct.

Danny



More information about the questions mailing list