[ntp:questions] Compensating for asymmetric delay on a per-peer/server basis?

Rich Wales richw at richw.org
Thu Sep 11 19:09:25 UTC 2014


> Did I miss something? The OP did not offer any evidence that there was
> path asymmetry, just that there was an unexplained offset between two
> GPS sync'd servers.

The asymmetry in this case is being caused by the characteristics of the
cable modem that connects my residence with the campus network and the
rest of the Internet.

I've observed this consistently for several years, at various times of
day.  I'm convinced that it is not being caused by traffic congestion
-- or, at least, that any traffic congestion factor is small compared
to a pretty constant offset of 2 - 3 msec.

> The asymmetry is caused by asymmetric latency which is caused (for
> our purposes) by asymmetric line speeds.  Traffic shaping can change
> various things (depending on where you do it) including effective line
> speed (packets/s not bits/s).  Perhaps shaping UDP is too hard.

I would certainly be open to doing experimentation in this regard.  I
have essentially full control over two Linux boxes (one on either side
of the cable modem).  However, up till now at least, I don't have any
experience at all with traffic shaping; suggested steps are welcome.

In the absence of a solution from the traffic-shaping domain, I would
like to be able to use the "fudge" command in conjunction with a peer
or server (currently, "fudge" works only with a reference clock).  It
seems to me that the "time1" parameter would do what I want -- if it
could be specified for a peer or server, which currently it cannot.

Again, I'm running version 4.2.6p5 right now.

Rich Wales
richw at richw.org


More information about the questions mailing list