[ntp:questions] demonstrate traceability to UTC

Ryan Malayter malayter at gmail.com
Thu Mar 12 22:00:39 UTC 2009

On Thu, Mar 12, 2009 at 11:06 AM, Tom <tom_e_tek at hotmail.com> wrote:
> To All,
> How do you demonstrate the traceability of the NTP solution to BIPM
> UTC?

For "hard" tracability, you really need your own reference clock
(typically GPS) that you control, along with several internet based
sources as a sanity check against it.

If you don't want a reference clock, but want "tracable" time from
internet-based sources, you need to use some form of NTP
authentication. Typically, this means sharing a "signing" key with at
least one (more are better) upstream servers run by organizations that
are known to provide good time tracable to UTC. The NTP Autokey
protocol was designed to do this without sharing secrets, but is not
widely used, and does not work if there is network address translation

Is this a regulatory requirement? If so, what nation/regulation are we
talking about? Some may require the use of a local reference clock of
some type.

In the USA, for eample, you may be able to get authenticated time
directly from NIST:


More information about the questions mailing list