Re: [Ntp] Symmetric mode

[David Venhoek <[e-mail]>]

In the interest of trying to at least get to some kind of conclusion,
I have tried my best to write up the main conclusions from all the
discussions:

Symmetric connection mode:
 - uses only 2 packets per polling interval to exchange time between 2
servers, compared to 4 when both are configured to use client mode
with the other.
 - can be easier to configure as only 1 of the two servers needs their
configuration altered (although this brings some security questions)
 - there may/are some differences in how timing data produced by these
connections is processed in the NTPd implementation

Broadcast connection mode:
 - uses only 1 packet per client per polling interval instead of 2 in
client-server mode, and 1 general broadcast packet
 - can be easier to configure as clients can just pick up the
broadcast packet without needing to know about the server in advance
(although this brings some security questions)


Primarily a question to Danny and/or Harlan, but is this a decent
description of the main differences between these modes and
Client/Server mode?

Kind regards,
David Venhoek

On Sun, Oct 2, 2022 at 11:09 AM Doug Arnold
<doug.arnold=[e-mail]> wrote:
>
> It seems to me that the over-the-wire protocol should only be about sending timestamps back and forth. What the timestamps are used for belongs in the algorithms and clock control subsystems.
>
>
>
> From: Hal Murray <[e-mail]>
> Date: Friday, September 30, 2022 at 10:30 PM
> To: Doug Arnold <[e-mail]>
> Cc: Hal Murray <[e-mail]>, [e-mail] <[e-mail]>
> Subject: Re: [Ntp] Symmetric mode
>
>
> [e-mail] said:
> > Please enlighten me. I've been reading the discussion on symmetric ntp, a
> > nd I don=92t understand the purpose. An ntp server can ask other servers for
> > time using client server ntp, and use that information to examine its own
> > time and that of the peer it is looking at.  Why is a different over-the-wire
> > version of the protocol needed?
>
> That's what we are trying to figure out.
>
> Miroslav's draft for NTPv5 dropped symmetric mode.  Some people think it is
> interesting but they haven't managed to convince some others of us that it is
> important.
>
>
>
> --
> These are my opinions.  I hate spam.
>
>
> _______________________________________________
> ntp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ntp