This is the record of the testing done on the AP using the LANforge.
Inside the anechoic chamber, the LANforge (eth1) is connected to the AP under test via a LAN cable.
I set up 4 virtual stations in the LANforge, with the same SSID as the AP. This results in a loop that allows the LANforge to measure the throughputs of the various streams. The stations are named:
STA-1-BK (background)
STA-2-BE (best effort)
STA-3-VI (video)
STA-4-VO (voice)
I configure 1 uplink stream per station. Each stream is set as 10 Mbps UDP. Notice that the streams are in an increasing order of priority.
Tests are conducted when WMM is disabled, and the same tests are conducted when WMM is enabled.
The test procedure is as follows:
STA-1 starts transmitting.
Then STA-i starts transmitting at time (i-1)*30 s.
I found that the video and voice streams are not able to achieve the desired 10 Mbps, when WMM is disabled.
Next, WMM is enabled.
When WMM is enabled, the QoS priority of the data streams are taken into account. When the video or voice streams are turned on, they maintain the constant desired throughput, while the background and best effort streams are starved of throughput.
Also, when WMM is enabled, only the background and best effort streams shows large latency. The video and voice streams have very little latency.
Therefore, it can be seen that the WMM feature is working in AP under test.
Reference:
Wi-Fi Multimedia (WMM)
www.uniroma2.it/didattica/TPI2/deposito/wmm.pdf
As Ron said, 3-4% packet loss is incredibly high. ThousandEyes ran some experiments on the effects of packet loss and found anything over ~0.5% is hugely detrimental. see: https://blog.thousandeyes.com/a-very-simple-model-for-tcp-throughput/
Hopefully replacing the fiber patch (make sure to clean it first!) will help. Otherwise start looking into other sources for the loss.
Best Answer
When more traffic than a link can handle is sent to a link, the interface to that link will end up dropping the packets destined for that link. This is congestion, and there are some small mitigations for congestion, such as queuing, but, when it comes to networks, packets are preferred to be dropped as soon as possible rather than delaying them.
In your case, the dropped traffic never makes it out of the interface on the device which is generating the traffic.