> From: Nigel Bovey <[EMAIL PROTECTED]> 
> Sent: Fri, 09 Jan 2004 07:28:34 +1300 
> 
> On Thu, 08 Jan 2004 21:33, you wrote:
> 
> > If the ping time is < 100ms the host _must_ be in NZ. Laws of
> > physics.
> 
> Very untrue.
> 
> [...]
> 
> At no point would ping be a reliable way to discover the physical
> location of an IP address.

Yes, it would be reliable, but only in providing an upper limit. I understood 
Christopher's argument as being based
on the signal propagation speed. That way, if you get a certain
reply latency, you have at least an upper limit for the distance.
Assuming either fibre-optic or coaxial cable, signals travel at
approximately 2*10^8 m/s. So a round-trip delay of 10 ms would
be equivalent to 1000 km distance. Just one 'zero' too much...

The trouble is that you normally have some equipment like routers
and switches on the way, and the latency of these does not only
depend on their speed and type, but also on the actual network load.
Plus, the latencies caused by active networking equipment can be
quite long, as compared to the cable latency, even more so for
geographically short connections.

But to get back to the original question: Would traceroute provide
any useful information? The the next question is: what does an ISP
actually mean with "national" and "international"? The target? Any
points on the route there? I remember regularly having inside-Germany
connections routed via the USA...

Could be a tricky question in a meshed network with dynamic routing.

Next question, then: would they bill you based on the routing at the
time when, say, a TCP connection was established? Or would they really
bill based on a per-IP-packet count? Must be real fun getting the
billing system working all right :-)

Kind regards,

Helmut.



Reply via email to