Inconsistent Wireless Speeds Across Access Point Models in the Network

[MIKI.H]

Hi,

We have encountered a strange issue in our network that is causing slowness.
We noticed that when we run the command netsh wlan show interface  in the CMD on a computer, we get different speeds depending on which wireless APs we are connected to.

Whenever we connect to wireless units of model 1852 (which are the units deployed at our site), we experience lower speeds compared to test units we brought in specifically for evaluation: one model 1832 and one model 9115.

Our WLC was a 5520, and we upgraded to a 9800, but we are still seeing the same issue.

We are unable to understand why this discrepancy is occurring.

[Flavio Miranda]

@MIKI.H 

 I wouldn´t rely on "netsh wlan show interface" to measure the wifi quality. Make tests with iPerf with one server in the wired network and a wireless client.

[MIKI.H]

I got different results
thank you

[Flavio Miranda]

@MIKI.H 

 There are indeed a considerable difference.

This test was ran from the same machine?

[MIKI.H]

Yes, this is from the same machine (laptop). In the screenshots, the first one is from the PC side, and the second one is from the server side (the server is physically connected to the network). We just switched between the APs.

In addition, if I’m not mistaken, if I have a network running at 1 Gbps and the APs operate on 20 MHz, I should expect bandwidth from the APs to be somewhere between 80–100 Mbps. Is that correct?
If the answer is yes, how did I manage to get 200 Mbps when I replaced the AP?

Thanks.

[MIKI.H]

I believe I know what happened, and it seems there’s no issue at all.

During our testing with the 1852 wireless unit, we achieved 130 Mbps, which is relatively good.
In the second test, we used a 9115 wireless unit that supports AX. Based on what I observed, this unit can significantly enhance performance and speeds.

As seen in the image, I was connected via 802.11ax, where we achieved high and excellent speeds.

Regarding the iPerf test: we attempted it again with the 1832 wireless unit and initially observed low performance. However, we noticed that our computer was connected to a more distant unit. After the nearby wireless unit fully came online and the computer connected to it, we observed some initial instability. Once it stabilized, we ran the iPerf test again and achieved the same results as with the 1852.

In conclusion:
Patience is key when testing wireless performance.

[Flavio Miranda]

Great info. We need to make sure the test scenario is fair or we can have low quality value results

[JPavonM]

The speed of the link is something variable and not subject to guessing as it depend on the AP that the connection is stablished to, but the negotiation between the AP and the wireless adapter on the device (sometimes the logic inside the driver has little logic), and the background noise.

[MIKI.H]

The test was conducted using the same equipment and distances, but with different wireless units. I disconnected the 1852 and connected the 9115, and the result was different.

thanks.

[Rich R]

1. As @JPavonM said the data rate (speed of the link) you observe in Windows (or any client) varies all the time. Noise/interference and other environmental factors will influence that so that you can see it change from one minute to the next even on the same client with the same AP.  A person walking between the AP and the device could be enough to cause such a change.

2. Make sure the Intel drivers are up to date - there are very well known major bugs in older versions of the drivers: https://www.intel.com/content/www/us/en/download/19351/intel-wireless-wi-fi-drivers-for-windows-10-and-windows-11.html

3. Take into account the different capabilities of those AP models.  1832 and 1852 are 802.11ac Wave 2, 9115 is 802.11ax (WiFi 6).
1832 - 3x3 MU-MIMO with 2 spatial streams
1852 - 4x4 MU-MIMO with 3 spatial streams
9115 - 4x4 MU-MIMO with 4 spatial streams + other ax enhancements including OFDMA

4. Theoretical maximum data rates can be derived from https://mcsindex.com/
> APs operate on 20 MHz ... how did I manage to get 200 Mbps when I replaced the AP?
As you can see from the tables higher data rates are quite possible.  Every evolution of the WiFi standards delivers higher data rates.

5. > Patience is key when testing wireless performance.
Remember DFS CAC wait time can take minutes before the AP can use a DFS channel on the 5GHz band.