Re: [Ntp] NTPv5: big picture

[Magnus Danielson <[e-mail]>]

Doug and Hal,

Indeed. You can do much much better in fact. Many times it's not the
protocol format that is the main limit, but other protocol features.

I know of designs which while not using NTPv4 use a propriate format, it
could very well have been NTPv4 format packets.

So, it's more than how NTP packets look, it's how typical
implementations are done with the implementation constraints that NTP
comes with. Release constraints and think differently, and you can get
far better quality.

While future NTP may not need to go into all the available design space,
as some is already taken by for instance PTP or proprietary protocols,
there is still good reasons to leave room for future enhancements.

Cheers,
Magnus

On 2021-01-04 17:05, Doug Arnold wrote:
> Separating the algorithms from the message formats, and sequence allows for more flexibility for niche applications.  I know of a company that successfully sells a proprietary non-standard variant of ntpv4 for just this purpose.  More accuracy can be achieved with on path support, higher message rates and different servo filters.  
>
> Doug
>
> On 1/1/21, 4:57 PM, "Hal Murray" <[e-mail]> wrote:
>
>
>     [e-mail] said:
>     > I would say that the unifying themes for ntpv5 are greater flexibility, and
>     > higher accuracy, compared to ntpv4.   
>
>     Do you have any ideas on how to improve accuracy?
>
>     What are you thinking of for flexibility?  There has been a lot of pressure to 
>     require feature X.  Examples are TAI-UTC offset and authentication.  That 
>     seems like reducing flexibility.
>
>     -- 
>     These are my opinions.  I hate spam.
>
>
>
>
> _______________________________________________
> ntp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ntp