[ntp:questions] Re: Good GPS for attic?

[David Schwartz]

"Danny Mayer" <mayer at gis.net> wrote in message 
news:mailman.117.1108360834.583.questions at lists.ntp.isc.org...

> True enough. I was merely pointing out that it was a day 1 design
> consideration rather than an afterthought or not considered at all.
>
> Danny

    The problem is, when you have a computer that's connected to the 
Internet, one of the most common reasons you want your clock synchronized is 
so that you can timestamp network events. This just will not work if your 
time synchronization is over the same network links you are trying to 
measure.

    Here's an example for you. You have two network sites, one in New York 
and one in California. You want to know if the New York to California route 
is faster or slower than the California to New York route. Would you think 
you could do this reliably by using Internet time synchronization?

    Here's another one, same situation. You have events generated around the 
world that are sent over various mechanisms, including over the Internet. 
You want to know, for many of these events, if they got to New York or 
California first. Again, would you think you could do this reliably using 
Internet time synchronization?

    Here's another one. You have sites all over the country, or maybe the 
world. They exchange data with each other in complex ways over the Internet. 
You need to piece their logs together to figure out what happened in what 
order.

    A local reference clock can provide guaranteed accuracy. +/- 3uS 99.5% 
of the time is not difficult to achieve or even expensive.

    DS