In part 2.1, I briefly explained the difference between bandwidth, throughput, and speed. In this post, I will set a baseline by comparing wired to wireless performance. But first, let’s understand what are we trying to measure?
Many times I get a screenshot like this one and the customer starts complaining that Wi-Fi is slow. My first reaction is this an Internet Speed test and not Wi-Fi performance test! What we want to measure is the Wi-Fi performance and not the internet performance. Internet performance will be impacted by many factors including your Internet speed, shaping, filtering and your actual internet link utilization.
Proper Wi-Fi needs Proper Wired Network!
To assess the performance of the wireless network at a particular location, it is essential to check the performance of its associated wired network. This can help us eliminate sources of problems from the wired side.It is important to have a clear understanding of the traffic path to be able to choose the right points for testing. After gathering info about the issue or requirements, as a basic check, I try to examine the speed, duplex, and PoE/PoE+ settings for the port where the Access Point is connected. I also check the errors on the interface to eliminate any layer 1 issues. I then try if possible to connect a wired device to the port where the AP is connected (or another port on the same switch with identical configuration) and try to test from there. I prefer to use my laptop as an iPerf client and I use another device or WlanPi as an IPerf server.Choosing the right location for the IPerf Server is key to understanding the performance of the network . As such, now I can have a baseline of the expected throughput on the wired network.
Wi-Fi is different from wired!
It is very critical to understand the Wi-Fi is not like wired! Wired connections are usually full-duplex connections while Wi-Fi connection is a half-duplex connection over a shared medium. In wireless, one device can talk at the same time on the same channel in a particular coverage area. There is a lot of overhead in Wi-Fi mechanism to avoid interference. In wired networks with switches, multiple clients can communicate back and forth at full speed without the need to contend for the medium.
After establishing the baseline from the wired network, I start checking the wireless network. Few things I usually check include:
- Client Capabilities – What 802.11technology does the client support? What is the largest MCS that it can support? At what MCS is it connected? What is the signal strength that I am getting? What is the SNR?
- AP Capabilities – What technology is the AP using? 802.11n? 802.11ac? 802.11ax? Is the AP using 20M or 40M or 80M channels? Tools like WinFi lite or WiFi Explorer Pro are great here to get information about Channel Status – Unfortunately, I don’t have a sidekick to look at the spectrum but I use the channel utilization from the above tools to give me a better insight
After that, I start testing using a WlanPi. I connect the WlanPi to the wired network and I test from the wireless client multiple times. This will give me an initial rough idea of the performance of the network at that particular time with those particular settings.
I then continue on checking other settings. But first, we need to understand the MCS table to understand our theoretical limits. This will be explained in the next post.