ZenWifi XT8 w/ Gnuton, weird Wifi Issues

[glens]

Considerably much, and consistently.

Forgot to mention they're running GNUton's Merlin, and, naturally, that the AP "node" is fed by wire. Everything-all-on-one-channel might have an upside some way, but in my use and for my expectations, letting the load be shared throughout the spectrum vastly enhances performance. Perhaps most particularly regarding inter-client traffic, but I'm confident any multiple client-to-Internet traffic flows better as well.

I've surmised this before, and I may still be wrong about it, but my gut tells me that Asus doesn't adequately take into consideration the spectrum environment(s) of the node(s) vs. that of the controlling unit when deciding channel assignment. As router / AP each unit has a functional web interface and allows for individual channel assignments, power levels, and wifi schemes (4, 5, 6, etc.).

Beyond the mixed blessing of individual control the only downside to not using mesh is the loss of propagating a system-wide VLAN via a few browser forms. Have not yet found a need for that here, but acknowledge the possible value loss.

 

[Morris]

Considerably much, and consistently.

Forgot to mention they're running GNUton's Merlin, and, naturally, that the AP "node" is fed by wire. Everything-all-on-one-channel might have an upside some way, but in my use and for my expectations, letting the load be shared throughout the spectrum vastly enhances performance. Perhaps most particularly regarding inter-client traffic, but I'm confident any multiple client-to-Internet traffic flows better as well.

I've surmised this before, and I may still be wrong about it, but my gut tells me that Asus doesn't adequately take into consideration the spectrum environment(s) of the node(s) vs. that of the controlling unit when deciding channel assignment. As router / AP each unit has a functional web interface and allows for individual channel assignments, power levels, and wifi schemes (4, 5, 6, etc.).

Beyond the mixed blessing of individual control the only downside to not using mesh is the loss of propagating a system-wide VLAN via a few browser forms. Have not yet found a need for that here, but acknowledge the possible value loss.

Thank you,

Your answer confirms my expectations. Running mesh, we get convenience; guest SSIDs on the nodes and leave some performance on the table. I have not seen anyone demonstrate how much performance we are lousing to cross channel interference. My guess is it's not significant except when one's network is very busy and/or in situations where performance is critical. Most home networks are not that demanding yet there are those that are. Have you run any measurements to see the difference?

 

[Tech9]

I have not seen anyone demonstrate how much performance we are lousing to cross channel interference.

I have run tests with AiMesh and AP Mode on different channels with 2x identical Asus routers. The aggregate throughput to wireless clients in AP Mode is up to 2x higher. On the same channels the APs see each other and wait to transmit/receive. Not interference, just how Wi-Fi works. On different channels they transmit/receive simultaneously.

I also run a setup with 8x Wi-Fi 5 APs on 5GHz @40MHz wide channels serving >100 wireless clients. They can do >1.2Gbps aggregate throughput in a commercial building with some other Wi-Fi networks around. If I run all APs on the same channels @80MHz they can't exceed 0.6Gbps aggregate no matter what. Distant APs still see each other.

 

[glens]

My testing has been much less extensive. Perhaps I could eventually approach that 2:1 ratio but believe it (the ratio) was somewhat greater than that for me.

If one primarily has a client or two browsing the WWW, or maybe streaming some entertainment via the WAN, I doubt much of AiMesh's downside ever gets noticed. Pick up the pace and scope of the wifi usage, though, and it just falls flat.

 

[Tech9]

Home "mesh" systems use the same channels - it is what it is. The more "nodes" or "satellites" you add the worse it gets. Most people believe more Wi-Fi with better signal strength is better. Marketing is involved too - companies around are leasing pluggable "pods" for a monthly fee. The more you "need" or order the more you pay.

 

[Morris]

Home "mesh" systems use the same channels - it is what it is. The more "nodes" or "satellites" you add the worse it gets. Most people believe more Wi-Fi with better signal strength is better. Marketing is involved too - companies around are leasing pluggable "pods" for a monthly fee. The more you "need" or order the more you pay.

I've stayed in inquiry as my experience is in college campus WiFi. 2.4 GHz is just a mess yet it's the only way to get reach into obscure locations and yes many AP's hear each other. Those that don't are where it is harder as now client collision detect and AP collision detection come into play. For 5-Ghz there are more channels to move to when you keep the channel width down. This is where it becomes an art if there is demand for throughput and not just service. I agree with you that Asus is big on marketing and there flood the shelves with options costs there developers and support organization a lot of time taking away from a simple improvement such as allowing us to manually set the channel and signal strength on the nodes. Even auto channel could work, after all the code for both manual and auto is already on the device. Adding this to the existing manual options for mesh nodes does not seem like a lot of work to me. I believe I've seen some brands mesh nodes do individual channels, I don't recall which ones. I'm looking forward to Asus implementing VLANs as this seems like it may be the fix. If Asus had 2 or 3 models for each WiFi standard, making theses changes would not be such a huge effort.

For my own network, I have a router and two nodes. The router covers my house yet I have WiFi needs outdoors and while 2.4 GHz reaches the outdoor locations from the center of the house router, throughput is meh outside. I added the two nodes to get 5 GHz outside and this works yet with all 3 on the same channel throughput inside is about half of what I had before adding the nodes. I've been living this way for the convince of mesh management and because I have a very reliable network. The only time I notice the throughput issue is when automated backups of our laptops run as they take twice as long as they should. As an engineer this frustrates me yet when I put my managers cap on I see happy end users. If I were to change the nodes to AP mode, then for the rare times that a client is not getting good service, I'd have to search for the client in multiple places, a different frustration. My gear is only a year or two old so I'll sit back for now and wait for my next upgrade and see where things are at then.

 

[Tech9]

This is where it becomes an art if there is demand for throughput and not just service.

Correct, Wi-Fi ensuring throughput to multiple clients needs planning.

I've been living this way for the convince

Convenience comes first with home mesh systems. AiMesh is no different.

 

[kenkyee]

What are your issues? Nobody can help you as we don't know what your problem(s) are.

Sorry...same as OP.

I.e., WiFi seems ok for a while and then it stutters/lags for a while, then goes back to normal. So I was wondering how to tell if I have v1 vs. v2 hardware. I also tried an ET8 pair with the same behavior.

I'll start up a separate thread because AI mesh behavior is a bit strange.

 

[Tech9]

I also tried an ET8 pair with the same behavior.

I told you already XT8 and ET8 are exactly the same hardware devices.

I was wondering how to tell if I have v1 vs. v2 hardware

The router has a label at the bottom. It will indicate HW Ver.: 2.0 for newer units.

 

[glens]

I'll have to look again if I remember next time I'm near one, but I'm thinking my v1 units were made in Vietnam. Shouldn't make any difference but perhaps it does. (maybe mine are v2 of v1 units, hahaha!)

[edit: yep, production year '22 Vietnam]

 

[SirLouen]

Have anyone tested with the latest GNUton firmwares? Still the same issue? I have been time constrain for a long time, and I 'm trying to free myself to do some further testing