[ntp:questions] ARRGH!!! I woke up to a 50 SECOND clock error.

unruh unruh at invalid.ca
Thu Mar 15 21:09:33 UTC 2012


On 2012-03-15, Chris Albertson <albertson.chris at gmail.com> wrote:
>> Basically you have a trade-off between frequency error and time offset.
>> IIRC, polling quickly will reduce the offset, but you get greater frequency
>> variations.
>
>
> There is nothing magic about NTP.   You have the exact same trade off
> if you try and sync your wrist watch to the big town clock on the
> tower.
>
> If you look at the tower every hour and set the wrist watch to match
> you will have very small offset between the watch and the big clock.
> But if you do this the _rate_ of your wrist watch will never be
> spot-on except by luck.
>
> On the other hand if you sync your watch once then wait a week or a
> month to see how much it has gained or lost you may see quite a bit of
> "offset" but now you have good data on the relative rates of the two
> clocks
>
> NTP did not invent this trade off, it existed 200 years ago.

Unfortunately it is not that simple. That rate changes by significan
amounts. Thus the rate you get after a week may be very different than
the rate you get after an hour. That, I submit, is the chief obstacle to
having an accurate clock. And that change in rate does not fit with the
"Allan variance" assumptions (the noise source is not of the type
assumed)


>
>
>
>
> Chris Albertson
> Redondo Beach, California



More information about the questions mailing list