[ntp:questions] Is the use of the leap seconds mechanism mandatory?

Steve Kostecke kostecke at ntp.org
Fri Feb 27 21:27:21 UTC 2009


On 2009-02-27, H. Peter Anvin <hpa at zytor.com> wrote:

> Steve Kostecke wrote:
>
>> On 2009-02-26, Bartholome, Alain <alain.bartholome at eads.com> wrote:
>>
>>> The primary reason why I need NTP is the synchronization of the
>>> systems, not the accuracy of time.
>>
>> NTP is designed to synchronize computer clocks to a common time base.
>>
>> The most commonly used timebase is UTC acquired over a network or via
>> a radio ref-clock such as GPS. It's cheap. It's ubiquitous.
>>
>> A nice side effect of using UTC as your timebase is that your clocks
>> will be set to the correct time.
>
> Sure, but there are definitely applications where TAI are preferred to
> UTC.

The OP did not state a preferrence for TAI over UTC.

The OP _did_ repeat the old misguided "all I care about is
synchronization" meme about NTP. This is frequently uttered by people
who want to operate a stable time island and want to have it for free.

NTP needs a stable time base / frequency reference / what ever you
want to call it to achieve the best possible clock stability. Without
one your clocks are like a herd of kittens playing follow the leader.

UTC is ubiquitous and cheap. A stand alone rubidium based PPS source
will cost you at least $1K + the cost of integration.

-- 
Steve Kostecke <kostecke at ntp.org>
NTP Public Services Project - http://support.ntp.org/




More information about the questions mailing list