> The problem is that there is no way for NTP servers to determine
> whether or to what degree a connection is asymmetric.  All they know
> is how much time elapses from the moment they send out a request and
> the moment they receive a reply.

> I'm not even sure it's safe to assume that with enough peers, the
> assymetry of the respective connections will cancel each other out.

Sometimes it won't, yes.  For example, if my pool host were on a
severely asymmetric line, my time would be shifted with respect to the
rest of the world.  But this would matter only if my host's time were
compared with the rest of the world's by some other channel, such as if
I had a stratum-1 time source or if I had a second link with different
latency characteristics.  When the shift is always in the same
direction by the same amount, asymmetric latency just means that the
host's time is shifted with respect to the rest of the world - but, as
seen through the asymmetric-latency link, it's shifted back into
correctness.

That's not what's going on here, though, or the pool monitoring would
be seeing correct time.

/~\ The ASCII                             Mouse
\ / Ribbon Campaign
 X  Against HTML                [email protected]
/ \ Email!           7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B
_______________________________________________
pool mailing list
[email protected]
http://lists.ntp.org/listinfo/pool

Reply via email to