[ntp:questions] Asymmetric Network and Delay Questions.

Matt Kinard kinard at Raytheon.com
Thu Jul 14 13:32:40 UTC 2005


What are the implications of running ntpdate out of the crontab in an asymmetric
case?

I know that in a "best case scenario" that ntpd will be significantly more accurate
than ntpdate but what about in a situation like this?

Thanks,
Matt

Brad Knowles wrote:

> At 10:46 AM +0200 2005-07-14, Brad Knowles wrote:
>
> >       If there is significant asymmetry, you can pretty much throw out all the
> >  NTP algorithms, because you are violating one of the most basic assumptions.
>
>         I should mention one other thing -- if you use a broadcast NTP
> service, then you don't have to worry about asymmetry.  The clients
> just sit and passively listen to the server broadcasting the time,
> and then apply what they receive.
>
>         However, if you want to correct for any latency, then you have to
> calculate an average latency value that should be applied to all
> calculations, although it might be possible to vary this average
> latency value on a client-by-client basis.  For example, you might
> have one NTP server that is broadcasting time sync values via
> satellite (in geosync orbit) to a wide variety of listening stations
> in embassies around the world, and you can separately calculate the
> average propagation time to each different embassy and apply that
> different latency value to the clients at that particular site.
>
>         This is not the best example, since what you should really do is
> use GPS with good stratum 1 receivers (designed for timekeeping) at
> each client site, but it does illustrate the point of applying
> different latency/propagation values for each set of clients, and
> basing that on a broadcast model.
>
> --
> Brad Knowles, <brad at stop.mail-abuse.org>
>
> "Those who would give up essential Liberty, to purchase a little
> temporary Safety, deserve neither Liberty nor Safety."
>
>      -- Benjamin Franklin (1706-1790), reply of the Pennsylvania
>      Assembly to the Governor, November 11, 1755
>
>    SAGE member since 1995.  See <http://www.sage.org/> for more info.



More information about the questions mailing list