On Wed, Mar 30, 2016 at 12:41:17PM +0200, Dieter Sibold wrote:
> > >
> > >I also assume that to be able to get a very accurate results. But
> > >if that's the case I think they're going about this the wrong way.
> > >I think for most cases the overhead of the crypto is not going to
> > >have an impact on the results because there is just way more
> > >jitter on the network. For the cases this matters both sides
> > >should take an additional timestamp when the packet was really sent,
> > >and have a follow up packet that contains the correct time.
> > 
> > This would be a very clean solution, but my impression was always that NTP 
> > did not offer follow-up type messages (as a reminder: for NTS, we try to 
> > not require modifications to the specification of the underlying time sync 
> > protocol). Am I wrong about his?
> > 
> This is correct. NTP as specified in RFC 5905 does not have the concept of a 
> follow up message, like for example PTP. I assume

I can't believe that there isn't a draft for this yet.  At least
David Mills has been talking about this years ago.

As far as I know, at least Linux and FreeBSD have support for
hardware timestamping of the packets, but it depends on the
hardware and driver support.


Kurt

_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc

Reply via email to