[Flent-users] SmokePing probe

Toke Høiland-Jørgensen toke at toke.dk
Sun Feb 18 21:16:25 EET 2018


Pete Heist <peteheist at gmail.com> writes:

>> On Feb 18, 2018, at 7:11 PM, Toke Høiland-Jørgensen <toke at toke.dk> wrote:
>> 
>> Pete Heist <peteheist at gmail.com> writes:
>> 
>>> The author pulled the probe in pretty quickly, so it’s already upstream. Some samples:
>>> 
>>> https://www.drhleny.cz/bufferbloat/smokeping/irtt_smokeping.png <https://www.drhleny.cz/bufferbloat/smokeping/irtt_smokeping.png>
>>> 
>>> https://www.drhleny.cz/bufferbloat/smokeping/irtt_smokeping2.png
>>> <https://www.drhleny.cz/bufferbloat/smokeping/irtt_smokeping2.png>
>> 
>> Awesome, nice work!
>> 
>> Are you using some kind of management interface for your 70 smokeping
>> instances, or is it all manual? :)
>
>
> It’s a single smokeping instance configured with 70 targets, divided
> by region. So it’s one host that does all the pinging to each of the
> target hosts. I didn’t configure it originally, so will be finding out
> more as I convince them to deploy irtt on at least some of the hosts
> for comparison with the ping results.
>
> Ultimately I’m trying to figure out what if any queues are building in
> the backhaul and if aqm would help there or not, however I can figure
> out how to do that. I hope to also try flent runs within the backhaul,
> but I’d like to be able to somehow measure how the queues are handling
> real world traffic as well...

The mtr tool might be useful for locating the bottlenecks (assuming
there are actually multiple l3 hops): http://www.bitwizard.nl/mtr/

-Toke




More information about the Flent-users mailing list