[ntp:questions] Keeping NTP Honest Redux

[Richard B. Gilbert]

Terje Mathisen wrote:
> Evandro Menezes wrote:
>> On Nov 6, 3:10 pm, Terje Mathisen<Terje.Mathi... at tmsw.no>  wrote:
>>>
>>> Making it more convenient to lock polling at 64s isn't a good idea, 
>>> imho.
>>
>> What do you suggest to mitigate the wild swings between daytime and
>> nighttime, sunny and cloudy, cold-front and warm-front, etc, of both
>> the offset and the frequency?
> 
> If this in fact a real problem, I'd suggest wrapping some insulation 
> around the clock crystal.
> 
> This would make temperature swings much slower, right?

If you REALLY care, use an Oven Controlled crystal oscillator. Most 
servers live in a data center in which the temperature is controlled to 
within +/- two degrees F.  I suspect that an OCXO will not buy you much 
in the way of greater stability but if you are scrambling in pursuit of 
the "one true time". . . .