Measuring wireless backhaul performance on XT8

[tbrown313]

I have just setup my ASUS XT8 system. I have the router in my basement and a node on the next floor up. Everything appears to be working fine and performance is good.

I am using wireless backhaul to connect the two nodes. My question is, how can I specifically measure the performance of the backhaul connection? The web interface to the router tells me that the node connection quality is "Great". Well, that's Great!, but what does it mean in real performance numbers? The manual says the LED will be white when the connection is strong. Hmm, maybe it is, maybe not. It looks kind of pink to me.

The node has two hardwired ethernet clients--a Tivo and a 4K Apple TV. I used Speedtest on the Apple TV and it does support my internet bandwidth of 100 mbits/second. But I want to verify that the backhaul handles higher bandwidths.

I have used iperf3 running a client on my iPad Pro, and a server on my Mac mini to measure wireless speeds around the house. The Mac mini is hardwired ethernet directly to the XT8 router in the basement. With the iPad next to the basement router, I get about 700 mbits/second. With the iPad upstairs next to the node, I get similar results.

However, how do I know that the iPad is connected to the node, rather than to the basement router? I think it is, but can I verify this somehow?

I have also though of taking my wife's iMac upstairs and connecting it to the node via ethernet. Then I know the connection and can test using iperf3.

Any built-in tools I haven't found to do this? Or other clever suggestions?

 

[SimoniZ]

iPerf3 is really good and if you really want to know how the wireless backhaul is performing you should test in both directions. For even better results, I would physically connect the devices used for the iPerf3 test to determine the maximum performance of the backhaul and rule out any performance loss caused by wireless connections to the specific AP.

As for how to tell whether your IPAD is connected to the main router or node, just go back to the AiMesh settings and the client option should already be selected. On the right side of the page, it will show all of the clients, either by name or the name will be the MAC address of the device.

The selected router/node is the one with the blue verticle bar to the left of the topology icon. Just click on the other node and check it's client list to see what is connected there.

On the client list you also have the option to bind a device to a specfic node by clicking on the chain icon; very helpful when testing.

 

[RogerSC]

The easiest way (for me) to tell what node a client is connected to is to just use the Asus Router smartphone app. The "Devices" tab on the bottom of the app brings up a page that lists all your clients according to the node to which they're connected. Easy for me, since I usually have my phone in my pocket, instead of having to go back to my desktop and bring up the web admin GUI to find out. Your considerations may be different, of course *smile*.

 

[tbrown313]

Thanks for the tips. I have done more testing and have some results.

My router node is connected to my cable modem via ethernet. I have an M1 Mac mini running Monterey connected via ethernet to the router node. This Mac is running the iperf3 server.

The second node is about 20 feet away, one floor above the router node, separated by just the floor/ceiling between the two floors.

I used an 2019 iPad Pro 12.9 running iperf3 in client mode for wireless testing in various parts of the house.The iPad has wireless AC.

I also have a 2017 Core i5 iMac with Big Sur running iperf3 in client mode that I used for ethernet testing.

For wired performance, I connected both the server and the iMac client to the router node via 1GB ethernet. The result was 940 Mbits/second.

To measure backhaul performance, I connected the iMac client to the remote node via 1GB ethernet. The result was 805 Mbits/second.

With the iPad wireless connected to the router node I got 655 Mbits/sec.

With the iPad connected via wireless to the remote node I got 594 Mbits/sec.

Part of the reason I was confused about performance through the backhaul was that I was not aware that devices take a few seconds at least, and maybe a minute or so, to switch from node to node. So, if I test with the iPad in the basement, then walk upstairs and test, the iPad tends to stay connected to the basement node. The difference in this case is not much. However, if I take the iPad to the farthest location in the house, probably 80+ feet away from the basement node, and through a ceiling/floor, plus 3-4 walls, it still stays connected to the basement node for a time. Then the performance drops to less than 50 Mbits/sec and occasionally to zero. Within a minute, it switches, and the performance improves dramatically going up to 300 Mbits/sec or so.

Overall, I am very pleased with the performance of the ZenWiFi XT8 system. I like the completeness of the Web GUl for administrationI. I am familiar with these types of interfaces for managing routers and they work well for me. The iPhone app is okay, but I only use that occasionally. I wish they had an iPad app. The iPhone app is stuck in portrait mode on my iPad, and given that I use a magic keyboard, that makes it pretty useless.

My previous router was a Netgear AC2600 which I had flashed with DDWRT. It worked well, but did not cover the far corners of the house. In addition, I would never recommend to anyone that they use DDWRT. I know enough about networking to get by, but DDWRT assumes/requires a lot more. Up to date documentation is spotty, and support is limited to their forums. If you manage to get the right person to respond to you, it's great, but it's a crapshoot. And releases are uncontrolled and without really good technical knowledge of WiFi, there is no way to know if a release will brick your router or muck up your wireless. That was a bad experiment. I was never able to get my smart thermostat to connect to the 2.4Ghz band using DDWRT, and the same was true of of my Bose Soundtouch speaker.

 

[SimoniZ]

I think your results look really good. Please keep in mind that the Ethernet Backhaul is maintaining your AiMesh setup, and any data sent by a device connected to your mesh node would also go over the Ethernet Backhaul, so with that overhead in mind, the 805 Mbps obtained during your testing looks good.

940 Mbps obtained with your wired test with devices connected to your main router is perfect (also overhead here as well).

Asus has a feature that isn't very popular on this forum called Smart Connect that will autuomatically move a device to the more preferred mesh node. I suspect that it's already enabled on your end, but if you do not like how long it takes for a device to switch to another node, you can adjust the RSSI values, but I would keep the default values or just adjust the current values by -5.

There is another option that should work well and just has one RSSI setting to adjust per band (2.4 GHz or 5 GHz) called Roaming assistant. Here you just input the value that is in dBm to tell the node when to move/disconnect a client from a specific AP.

Roaming assistant option is located under Wireless -> Advance and is usually set to -70 dBm. You can adjust this lower if you want a client to move earlier, like -60, -55, -50, etc. Please keep in mind that you do not want to go too low in dBm or else your wireless clients will be changing nodes frequently. Try -65 and if that isn't good enough, then move in -5 dBm increments.

I 'm happy your setup is working well and your test results look good!