ZenWiFi XT8: TriBand SmartConnect Faster than 5GHz SmartConnect?

[GBear]

So I finally got around to upgrading my XT8 mesh firmware (42095 to 46980) and did the recommended factory reset. While reconfiguring, I couldn't remember where the "Enable Ethernet Backhaul" setting was for my wired pair, so I found myself temporarily with a single SSID and Tri-Band SmartConnect. While there, I did a Speednet test on my laptop and got a "whoa" download speed of 750+ Mbps. Shortly after that I re-discovered the ethernet backhaul button and set up a separate 2.4GHz SSID (which in the past had been necessary for some devices and for the moment seemed easier than reconfiguring all the IoT devices to a new single SSID and seeing if they'll work). After that I was able set up an SSID with 5GHz SmartConnect on the two 5GHz bands like I had before.

I was surprised to run Speednet (really not very long after the previous test) and get a much lower throughput on my laptop: first 150 Mbps and then, after disconnecting and reconnecting because the result was slower than on my phone, about 400 Mbps on both devices. This is well below the 750 Mbps with the Tri-Band test.

I realize that Speednet isn't a definitive network test and that it's also possible that my WAN connection got more congested. And I'm not facile enough with the settings to just bounce back and forth doing more testing without possibly breaking something more important. Still, I'm pretty sure I didn't change any other settings so I'm wondering if there's anything intrinsic to Tri-Band vs. 5GHz SmartConnect that could explain the difference?

 

[tgl]

You don't say what sort of speed test results you have typically gotten in the past. Is 750 unusually high? Is 400 unusually low?

In my (admittedly limited) experience with speedtest.net since I got 1GB fiber service, it's painfully obvious that the bottleneck is often somewhere out in the network, because I can get wildly different results at different times even though local activity level is the same.

 

[GBear]

You don't say what sort of speed test results you have typically gotten in the past. Is 750 unusually high? Is 400 unusually low?

In my (admittedly limited) experience with speedtest.net since I got 1GB fiber service, it's painfully obvious that the bottleneck is often somewhere out in the network, because I can get wildly different results at different times even though local activity level is the same.

I hear you. I'd usually get around 600 Mbps on a good day and of course sometimes slower. Enough that 750 seemed like I may have made a great decision to upgrade the firmware before 400 brought me back to earth. As you say, maybe it's just the network, but such disparate results within an hour made me wonder if it might actually be my settings. Meanwhile, subsequent tests through right now have remained around 400.

(And I apologize for not getting Speedtest.net correct in my OP.)

 

[tgl]

I hear you. I'd usually get around 600 Mbps on a good day and of course sometimes slower. Enough that 750 seemed like I may have made a great decision to upgrade the firmware before 400 brought me back to earth. As you say, maybe it's just the network, but such disparate results within an hour made me wonder if it might actually be my settings. Meanwhile, subsequent tests through right now have remained around 400.

Hmm. The one 750 result is tantalizing, isn't it? Suggests that maybe there's some removable bottleneck you haven't recognized. Seems like it'd be worth poking at it harder. Can you try other devices or vary the test conditions?

I can't offer a lot of advice, but I can share my own experience with this. I got Verizon FiOS 1GB service last year, and what I find is that Verizon's own local speedtest.net server seems to be underprovisioned: it seldom gives me speeds approaching 1GB, and frequently quite a lot less (and quite asymmetrical up/down). However, there's also a nearby Comcast-hosted speedtest.net server, and when I use that one I can often measure 930-ish Mbps up and down speeds when testing from machines wired directly to the FiOS router. That of course is "GB" ethernet speed, so the bottleneck is the ethernet connections. I conclude that the fiber and the router are delivering the advertised speed, and when I see something less than that the bottleneck is somewhere else.

The reason I've become a SNB reader is my struggles to bring my house network on par with the FiOS connection :-(. Suffice it to say that GB-speed WiFi is a whole lot harder to achieve than the manufacturers would like you to believe, unless maybe you live in a thin-walled shack miles from any other radio users.