>Basically, a valid SNTP server is supposed to be sourced directly from >a reference, never from a remote NTP server.
Just to clarify: the recommendation is that the SNTP server get it directly from a reference in preference to gettting it from a remote server via SNTP (That remote server may or may not be an NTP server as well). But if the SNTP server gets it from a remote NTP server via NTP, that's fine. With both SNTP and NTP, it's always better to get time from a radio clock rather than a remote server. But if you set up a multilevel SNTP tree, the leaves are going to be less accurate than in an NTP tree of the same depth. I.e. depth costs you more in SNTP than in NTP. I think you have to take recommendations like this in the RFC with a grain of salt, because the authors don't know what your requirements are. Maybe <500 ms accuracy isn't worth the price of either a radio clock or NTP to you. In that case, a 3-deep SNTP network could be the best way to go. -- Bryan Henderson Phone 408-621-2000 San Jose, California _______________________________________________ questions mailing list [email protected] https://lists.ntp.isc.org/mailman/listinfo/questions
