Manav and Shahram What I proposed is exactly NOT updating NTP, but making a generic mechanism that MAY be used for ANY protocol that cares about time.
For NTP, all one needs to do is to give the present code corrected arrival times, that will automatically reduce PDV and thus increase accuracy. For RFC-868 the ToD is delivered with much reduced delay error. Where change is needed is for one-way delay measurement (e.g., OWAMP). Regarding updating NTP to a new generation, that is precisely one of TICTOC's charter items (and indeed in some people's eyes the only thing we should be doing). TICTOC's purpose is to find improvements to both NTP and 1588, certainly not only 1588 (which is an IEEE protocol, not an IETF one). Y(J)S -----Original Message----- From: Bhatia, Manav (Manav) [mailto:[email protected]] Sent: Thursday, February 03, 2011 01:31 To: [email protected]; Shahram Davari Cc: Yaakov Stein; [email protected] Subject: RE: [TICTOC] Transporting PTP messages (1588) over MPLS Networks Stewart, Shouldn't the work on retrofitting TC to NTP happen in the NTP WG? This sounds like NTPv4++ to me. Also as Shahram has earlier said, if we do decide to extend NTP for TC, then all that needs to change in the current draft is that the RSVP object will now point to the start of the NTP header instead of the PTP header. Most of the things defined in this draft should work for NTP as well. Cheers, Manav > -----Original Message----- > From: Stewart Bryant [mailto:[email protected]] > Sent: Wednesday, February 02, 2011 11.50 PM > To: Shahram Davari > Cc: Yaakov Stein; Bhatia, Manav (Manav); [email protected] > Subject: Re: [TICTOC] Transporting PTP messages (1588) over > MPLS Networks > > Shahram > > I think that it would be useful to retrofit TC to NTP to give > it greater > accuracy. > > There are far more NTP clients out there than there 1588 clients, a > situation that is likely to continue for the foreseeable future. > > Stewart > > > On 02/02/2011 18:02, Shahram Davari wrote: > > Hi Yaakov, > > > > But I thought NTP doesn't require TC capability (Correction > Field Update) at every hop. If so then why do we need such > capability for NTP? If what you have in mind is for faster > synchronization without waiting for local clock convergence > then that is not the way NTP is implemented today. > > > > Thx > > Shahram > > _______________________________________________ TICTOC mailing list [email protected] https://www.ietf.org/mailman/listinfo/tictoc
