Richard B. Gilbert <rgilber...@comcast.net> wrote: >> b) What are the criteria to consider in deciding when ntp pool project >> is enough for our needs ? > > 1. Availability > 2. Reliability > 3. Distance from your site. Absurd example: If you are in New York > City, you would NOT want to configure a server in Tokyo!
In fact, that (3) would be an advantage of the pool. When you would use the pool (and especially when you omit the obsolete zone names) you would not get a server that far away. When you configure it statically, it is your own responsibility to find where it is. > Don't use two servers! It is written that a man with two clocks can > never be certain what time it is. Four servers is generally regarded as > the minimum. A configuration of five servers can survive the failure of > two servers and a configuration of seven servers is able to survive the > failure of three. This chat comes up all the time, but it is not realistic. When you have two GPS receivers and the driver software monitors their reception status (as opposed to blindly assuming they are locked), having two receivers provides redundancy against failure of one. There is no issue with "two clocks", because both receivers are synchronized to the same clock and both times will be equal or very close. It does not matter to which of them you sync. Having 4 or 5 of such clocks is completely unreasonable. When you would want to protect against the GPS system going haywire, you would need to have clock receivers of different technology (and you would have to be sure that your alternate technology is not itself synchronizing to GPS). But every redundant or failsafe system has limitations in the errors it can correct, and recommending a user to buy 4 clocks because else there would be problems with his time service is generally not going to bring anything. _______________________________________________ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions