[ntp:questions] Re: Small fixed offset

Tom Smith smith at cag.zko.hp.com
Fri Oct 28 13:22:20 UTC 2005


ariel.burbaickij at gmail.com wrote:
> There is misunderstanding here.
> The signals are timestamped at the equipement after they are feed in
> into it.
> In order to get accurate timestamps (where by accurate I mean time as
> the
> signals where seen/generated in USA not as signals were feed in into
> equipment in Central Europe) we have to introduce small offset on the
> clocks
> on the equipment (they must run 150 milliseconds behind the real time)
> . It is not
> possible to timestamp signals in USA as the backhauling equipement is
> not much more than optical fibre cables.  Clocks
> are synchronized from private ntp server. I am not talking about
> compensating
> the network latency in NTP protocol itself. Is it more clear now?
> 

I'm still not quite sure what you are describing.

Is there an NTP server in the U.S. to which a GPS clock is attached
and NTP clients in Central Europe that use this NTP server? If so,
NTP itself takes care of removing any network latency from the
timestamps so that, in theory, the time on the NTP server in the
U.S. is *exactly* the same as the time on the NTP client in Central
Europe. This is determined dynamically, depending on the network
conditions existing at the time each poll is taken. It is not a
perfect calculation in that if the network has a larger delay
in one direction than in the other, there will be an error of
1/2 the delay difference. However, if your link is as tightly
specified as you imply, there should be little or no error in
this calculation.

If you instead mean that there is a GPS clock in the U.S. connected
directly in some fashion to an NTP server in Central Europe, then,
although I have no idea how or why you would do that, the latency
of the connection would be specified in the fudge statement for that
refclock.

-Tom




More information about the questions mailing list