this is my first time using Flent test, but am not sure if am doing it right or wrong, it seems that i would get almost the same results with or without SQM, am using my PC which is where Flent is installed on, i use the PC IP address for the test. i have Ubuntu installed as a operating system. 
You should test real-life connectivity into a remote host that has netperf running. Typically I use a site from a SQM developer: netperf-eu.bufferbloat.net
(earlier there were also US-based sites netperf.bufferbloat.net (also called netperf-east.bufferbloat.net) and netperf-west.bufferbloat.net , but I use the European one.)
Main usage of flent is to compare the latency and speed stability under load, so that you can see how QoS tools (like SQM) help managing the latency.
QoS tools' main goal is to maintain responsivity of the connection, so that one heavy data stream does not consume all others' bandwidth.
I use a script for running flent (in Ubuntu 18.04):
#!/bin/sh
# args: Device qdisc down up wifi/wired/etc
IDstring=$1_$2_$3_$4_$5
ping -c 10 netperf-eu.bufferbloat.net
date
echo flent ipv4 150s netperf-eu.bufferbloat.net $IDstring
flent --ipv4 -l 150 -H netperf-eu.bufferbloat.net rrul -p all_scaled -D . -t $IDstring -o $IDstring.png
date
The easiest chart to see an overview, is "all_scaled". It shows refined image of download and upload speeds, as well as the ping latency during the test.
Two images below:
- unlimited speed, SQM turned off. See how the download speed components fluctuate wildly and latency is over 50ms.
- SQM simple, speed limits down 105M and up 11M (suitable for my ISP) and latency is nicely kept at 21ms by SQM
Note that in both scenarios, the download speed was approx. 100 Mbit (4 x 25 Mbit of 4 streams) that I get from my ISP.
But SQM limited upload speed according to the upload limit I had set, so latency could be kept reasonable. (In the "unlimited" case, the upload speed was higher, but that choked the connection and thus increased latency.)
Unlimited
105M/11M
1 Like