Danny Mayer wrote:
I'm copying Dave Mills on this, maybe he can say more and that's also
the reason that I'm top-posting. . My feeling on this is that is a very
restricted set of possible problems and is unlikely to be useful for NTP
since most packets go through multiple hops and you are unlikely to get
the outbound path and the return path follow the same paths.
URL: http://www.owlnet.rice.edu/~gurewitz/4.pdf
Danny
You are correct that in the general case this will not work. However I
would not write it off so easily. There are subnetworks where this would
prove useful and which are either sufficiently constrained, or where the
TWTT packets could be sufficiently constrained, that this would work.
This is a good example of how applying both time transfer and routing
to the problem can produce a better solution than can be achieved
without both sets of expertise.
Just because we can't improve the time transfer for all NTP clients does
not mean that we should not improve it for some.
- Stewart
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc
