On Tue, Apr 29, 2014, at 20:21, James Bensley wrote:
> So to calculate the delay I can say (c in meters per second);
> 
> (c*0.6) * (31000*1.4) =
> (300,000,000m/s*0.6) * (180000000m) =
> 0.00024111111 seconds, or 0.24ms.

RTT = d1 + d2 + 2 * (ds + l / v)

Where:
d1, d2 = delays due to equipment (you can consider it zero, but be ready
to have bad surprises)
ds = serialisation delay. For 10Gbps and reasonably sized packets, you
can consider it zero or close to.
l = fiber length (happened to have 2 x "radial distance" in London)
v = c / 1.62 (for fiber)
2 - because you mentioned RTT, not one-way.

Your result is pretty realistic for one-way. The formula.... much less
so (the math too).

> Before this project could have even come to light if someone had said
> to me "Can you provide a link between these two PoPs that will have a
> latency lower than 2ms?" - Assuming all the above accurately relates
> to the real fibre properties I could say "Yes!" (subject to surveys,

Beware of the customer :)

Reply via email to