[time-nuts] Re: how much is my router influencing time-keeping over the network

[Keelan Lightfoot]

Smokeping is a favourite:

https://oss.oetiker.ch/smokeping/

Internet latency is my day job!

RTT measurements can be very noisy and unpredictable, and often you'll see
a pattern with a 24 hour period that reflects load patterns on the
internet. Also, the protocol does make a difference. TCP is basically
unusable for the purpose because of retransmits. ICMP and UDP are
soliciting different layers of the IP stack with different priorities, so
the results can be better or worse than expected. Thanks to physics, the
lower bound is much more predictable than the upper bound.

The "speed of light in the internet", is quite slow compared to the ideal
speed of light in a glass fiber. I should share some data I have that
characterizes the number on a global scale. Unsurprisingly, the speed of
light under the oceans is faster than the speed of light over land.

- Keelan


On Wed, Sep 14, 2022 at 12:14 PM folkert via time-nuts <
time-nuts at lists.febo.com> wrote:

> Hi,
>
> For fun I'm developing a router for HAM packet networks. What it does is
> route AX.25 packets between radios and tunnels (it can also bridge- and
> filter them).
>
> What I would like to measure now is, how bad does it influence
> time-keeping when syncing time takes place over a network. I could of
> course just setup tcp/ip and let two ntp instances sync over it and then
> calculate an allan deviation plot. But are there also ways that are more
> interesting or simpler? Something like a ping-plot? (hmmm, maybe I can
> do an Allan-D. plot from ping as well; measure send a ping and write
> down when the response came in)
>
> In all cases I would first measure things between my laptop and a
> virtual machine and if that works (regular ping works fine) then I'll do
> it over LoRa (a real radio connection).
>
> some ping statistics over a virtual lan over my router:
>
> 5824 packets transmitted, 5824 received, 0% packet loss, time 1459943ms
> rtt min/avg/max/mdev = 22.011/33.290/66.959/7.361 ms
>
>
> regards,
>
> PD9FVH (Dutch callsign)
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com
>