[Nottingham] "ntpq -p" displayed flags

Martin martin at ml1.co.uk
Fri Jan 27 17:30:59 UTC 2012

On 27/01/12 16:52, Michael Quaintance wrote:
>     > Will you be writing up PTP (IEEE 1588) next? ;-)
>     No.
>     In any case, there are various other peering systems appearing that are
>     fully distributed that the MPAA and RIAA cannot so easily vandalise.
> Martin, it seems you misunderstood.

Indeed so it seems. Just sticky context switching after the SOPA/PIPA
silliness... :-(

> PTP (IEEE 1588) is the Precision Time
> Protocol http://www.nist.gov/el/isd/ieee/ieee1588.cfm
> "This standard defines a protocol enabling precise synchronization of
> clocks in measurement and control systems implemented with technologies
> such as network communication, local computing and distributed objects.
> The protocol will be applicable to systems communicating by local area
> networks supporting multicast messaging including but not limited to
> Ethernet. The protocol will enable heterogeneous systems that include
> clocks of various inherent precision, resolution and stability to
> synchronize. The protocol will support system-wide synchronization
> accuracy in the sub-microsecond range with minimal network and local
> clock computing resources. The default behavior of the protocol will
> allow simple systems to be installed and operated without requiring the
> administrative attention of users."
> PTP is basically trying to be a better (more accurate) NTP. It's not
> Peer-to-Peer as you seem to have read it.

Mmmm... Not "Peer-to-Peer" ( ;-) ) at all in that it is a master-slave
synchronisation system for a LAN, rather than a peering synchronisation
as with NTP with stratums operating across a WAN...

See the diagram on http://www.nist.gov/el/isd/ieee/ieee1588.cfm

A more readable explanation is on:

And a software demon is:

Interesting if it is simpler and faster. It looks to be less robust than
NTP but then again it is only intended for LAN use.

Which comes back to the question of what accuracies can be achieved...

The claim for PTP is in microseconds. I just wonder if NTP achieves
similar results for a cluster synchronised across a LAN?

NTP synchronises to UTC (with leap seconds). PTP synchronises to a local
master clock for International Atomic Time (TAI, monotonic, no leaps).

Interesting, thanks. Another note and reference for the page!

Still looking for good comment on how to assess the accuracy/error of
NTP... What values report that and what can be believed?


Martin Lomas
martin at ml1.co.uk

More information about the Nottingham mailing list