On Mon, Apr 24, 2017 at 7:09 PM, Bill Unruh <un...@physics.ubc.ca> wrote:
> But what is a "reasonable network"? Is it defined as one where that does not
> happen? And how do you know your network is "reasonable"?

I think I used a few too many "reasonable's" in my descriptions. :)

I think what I was trying to say was that if you have a network that
is completely under your control then you can potentially make some
statements about what you think may or may not be likely in terms of
delay introduced at various points.

> I would not say it is unlikely. There may well be a reason why one leg or
> the other has a very consistant extra delay, which does not vary.

Yea, agreed.  I think this comes back to what I was saying above.  If
you control the whole thing, then perhaps you can make some stronger
statements about what may or may not be likely based on things like
hardware, software, data flows, etc.

> Sure you can. PUt other clocks which do not have that assymmetry on both
> ends.

Yes, of course.  I think I originally wrote something along the lines
"... unless you put other reference clocks on both ends" but then
decided it wasn't worth the extra verbiage.  Apologies for not being

> These statements are so wishy washy that they do not really add anything for
> anyone reading them.

Haha, nothing worse than being wishy washy.  I was just attempting to
state that I that it would be unlikely for the delays to change in
lock step like that, but I guess you could come up with plenty of
scenarios where this happens.  Again, coming back to what sorts of
claims you're willing to make about the network.

> See for example the graphs in www.theory.physics.ubc.ca/chrony/chrony.html
> at the bottom of the page. These are offset vs delay graphs. They all
> clearly
> show that the larger offsets come from assymetric random delays. There is a
> correlation between offset and delay which would not be there if the delays
> were symmetric and random. This of course does not mean that there are not
> also systematic assymmetry in
> the delays which would not show up in such a scatter plot.

Awesome, thanks!  I'll take a look at that page in more detail.

To unsubscribe email chrony-users-requ...@chrony.tuxfamily.org 
with "unsubscribe" in the subject.
For help email chrony-users-requ...@chrony.tuxfamily.org 
with "help" in the subject.
Trouble?  Email listmas...@chrony.tuxfamily.org.

Reply via email to