2 ms·
Cool of them to publish their testing methodologies, I agree with them very well. I just got done designing 3 wireless testing cages for the company I work for
by zamadatix 7y ago
Cool of them to publish their testing methodologies, I agree with them very well. I just got done designing 3 wireless testing cages for the company I work for so I've been looking a lot into what makes good testing methodology for wireless. Very nice to see them dive into airtime right away and they actually cover it pretty well. I wouldn't extrapolate their conclusions on mesh and why it's fine to anything beyond low density home environments but that's what 95% of the readers fit into so perfectly reasonable.
One grave conflation they make is they keep referring to web streaming as actually performing like a stream of data. In reality web streaming (outside of WebRTC) is actually very bursty chunks being sent usually in the range of every 1-5 seconds. In the Wi-Fi world the two types of streams can behave very differently. Not that their test is bad but it isn't necessarily testing what they say it represents.
The HTTP workload is pretty representitive of what it says it's testing. The rise of HTTP/2 changes the way 16 connections to a single site would work but in reality loading a single site often connects to multiple servers anyways.
I was very happy to see them point out the large swings in jitter/latency under load as well!
Unless you're in a rural area and don't have too many legacy wireless devices I'd really caution putting anything in 2.4 that doesn't need to be there. It does penetrate farther but it's also only 20 MHz of air space that almost always has to operate with legacy compatibility (damn you Nintendo devices and cheap IoT wireless N chips). If possible try to keep everything new on 5 GHz only and if you have a lot of devices or use a mesh utilize 2 40 or 80 MHz 5GHz channels depending on your wireless environment.
Overall very good and much better testing than I expected. If there were anything I'd add to such an article it'd be a section in any wireless testing article pointing out "you'll almost always be able to get better improvements by replacing your clients instead of comparing wireless routers of the same generation". Obviously more harder to do/more expensive but I think a lot of the time home users think the access point is the be all end all of how good your wireless will be, forgetting it's only 1 device out of a dozen in the house and just knowing this can set more realistic expectations and a better understanding of what you're actually trying to improve.