Chuck Swiger wrote:
> Tony Hoyle wrote:
> [ ... ]
>   
>> I've traditionally used the mcc.ac.uk ones (0.023 jitter)... I've got
>> about 40 servers on the list at the moment... I'll cut out the ones with
>> the bad jitter/delay then.  Even if I get a GPS source working I'd
>> rather know how far I am from reality.. if it gets too far I know
>> something is fubar.
>>     
>
> Normally, one should only pick perhaps three remote NTP servers, and set up a
> ring of three NTP peers locally.  Trying to talk to ~40 remote servers is
> probably not helpful.
>
>   
Actually you need at least four to reliably detect a bad server (3+n to
detect n bad servers) - if you support many clients it's a good idea to
have at least five to allow foe two upstream failures.

John

>   

_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to