XT8 with 1gb fiber issues

[tbroome81]

We just recently switched from cable internet to fiber. We have 4 XT8 routers (1 main, 3 nodes). Ethernet backhaul isn't really an option given the construction of our house. We have 25 smart devices (Ring cameras, MyQ garage door control, Amazon Echos, smart switches/plugs/etc), all on the 2.4GHZ band. Our Wi-Fi 6 devices are on the first 5GHZ band. We have 7 of those (Apple TV 4K, 2 MacBook Pros, 2 iPhone 13 Pros, 2 iPad Air 5th Gens). I also have 4 devices that are plugged into ethernet. At the main, the speed is near 1gb. In the living room (20 feet away, but with a node within 15 feet and a node within 6 feet), I can only pull AT MOST 400mb to 450mb on 5GHZ channel 1. Is this normal or am I missing something in configuration? Also, why won't ASUS allow the backhaul channel be chosen instead of it automatically being the one with the most theoretical throughput? Thanks!

 

[tbroome81]

We just recently switched from cable internet to fiber. We have 4 XT8 routers (1 main, 3 nodes). Ethernet backhaul isn't really an option given the construction of our house. We have 25 smart devices (Ring cameras, MyQ garage door control, Amazon Echos, smart switches/plugs/etc), all on the 2.4GHZ band. Our Wi-Fi 6 devices are on the first 5GHZ band. We have 7 of those (Apple TV 4K, 2 MacBook Pros, 2 iPhone 13 Pros, 2 iPad Air 5th Gens). I also have 4 devices that are plugged into ethernet. At the main, the speed is near 1gb. In the living room (20 feet away, but with a node within 15 feet and a node within 6 feet), I can only pull AT MOST 400mb to 450mb on 5GHZ channel 1. Is this normal or am I missing something in configuration? Also, why won't ASUS allow the backhaul channel be chosen instead of it automatically being the one with the most theoretical throughput? Thanks!

Update: My 2021 MacBook Pro is only pulling just below 200mb on Wi-Fi 6 this morning. I'm at a loss.

 

[OzarkEdge]

Update: My 2021 MacBook Pro is only pulling just below 200mb on Wi-Fi 6 this morning. I'm at a loss.

Find your client and node connection details in the Wireless Log to better know the strength of these wireless connections. Then you have better idea what to expect or try to improve.

MAC address
RSSI received signal strength
PHY wireless mode
NSS number of spacial streams
BW bandwidth
Tx transmit rate when actually transmitting (WiFi is a shared medium... every transmission takes its turn in time, one direction at a time)

What is the level area in sq ft and number of levels you are covering with 4 APs?

OE

 

[tbroome81]

We are covering an area just over 1000 square feet including the garage, but the signal has to penetrate through a double brick outer wall (brick over cinder block) for 4 of the smart devices.

When I look at the info on my MacBook Pro, here's what I get:

RSSI received signal strength: -40 dBm, -89 dBm Noise
PHY wireless mode: 802.11ax
NSS number of spacial streams: 2
Tx transmit rate: 720 mb/s

 

[tgl]

4 nodes for 1000sqft?? You've got too much wifi, methinks. One XT8 would do fine for your inside-the-house devices, at least if you can place it reasonably centrally. Or you could use two XT8s -- no more -- placed on opposite sides of the house but with (if you can't do wired backhaul) clear line of sight between them. With what you've got, you're inviting problems with unnecessary roaming between the nodes and lots of retransmissions.

Not sure about what to recommend for the outside-the-house devices; you did not give enough detail. Are they all on one side of the house, or scattered around? How far away? How much bandwidth do they need? If the reason you have so many XT8s is otherwise you can't get 2.4GHz signal to all those devices, what you might try is turning off the 5GHz-1 radio in all but the main node and maybe one other one (as per the alternatives I just gave). Spreading the 2.4GHz network around doesn't mean you have to do exactly the same with the 5GHz network.

 

[OzarkEdge]

We are covering an area just over 1000 square feet including the garage, but the signal has to penetrate through a double brick outer wall (brick over cinder block) for 4 of the smart devices.

When I look at the info on my MacBook Pro, here's what I get:

RSSI received signal strength: -40 dBm, -89 dBm Noise
PHY wireless mode: 802.11ax
NSS number of spacial streams: 2
Tx transmit rate: 720 mb/s

Probably too much WiFi.

Do you know how to login to the router and inspect its Wireless Log?

OE

 

[tgl]

When I look at the info on my MacBook Pro, here's what I get:

RSSI received signal strength: -40 dBm, -89 dBm Noise
PHY wireless mode: 802.11ax
NSS number of spacial streams: 2
Tx transmit rate: 720 mb/s

Looking closer at those numbers, 720 Mbps is pretty awful if you have signal as good as -40dBm. For comparison, my own MacBook Pro is currently showing me Tx rate 720 Mbps or better (it moves around some) at RSSI of -58dBm. The AP is about 25ft away on another floor of my house, so poor signal is expectable ... but if I were close enough to it to get RSSI -40, I'd expect the Tx rate to easily max out at 1200 Mbps. I've gotten that at RSSI -45 or worse.

The most obvious explanation for your poorer showing is interference, ie too much wifi (too many nodes nearby). As @OzarkEdge suggests, the XT8s' wireless log data might help diagnose that. But personally I'd start by just shutting down all but the main node and seeing what sort of performance you get with only that one. Then add just the minimum number of nodes carefully placed to cover any dead spots. Less is more.

 

[tbroome81]

I removed 2 nodes and speeds increased.

Now I have another question. Due to various constraints, I can't run a true ethernet backhaul to free up the 2nd 5GHZ channel. Could I use gigabit powerline adapters to accomplish the same thing, albeit a bit slower than if it was true ethernet backhaul?

 

[OzarkEdge]

I removed 2 nodes and speeds increased.

Now there's a tip you can take to the bank!

Now I have another question. Due to various constraints, I can't run a true ethernet backhaul to free up the 2nd 5GHZ channel band 5-2. Could I use gigabit powerline adapters to accomplish the same thing, albeit a bit slower than if it was true ethernet backhaul?

Yes, you can try powerline, subject to the usual concerns plus any AiMesh quirkiness, if any.

OE

 

[tgl]

Could I use gigabit powerline adapters to accomplish the same thing, albeit a bit slower than if it was true ethernet backhaul?

My own experience with powerline was pretty negative, but that's just one data point. If your house has modern up-to-code wiring, it might work okay --- especially if you can place the powerline adapters on the same circuit. Personally, I'd bet against it being fully satisfactory, so you might want to be careful to get the adapters from someplace with a good returns policy.

 

[kildare]

Could I use gigabit powerline adapters to accomplish the same thing, albeit a bit slower than if it was true ethernet backhaul?

What should be the benefit of having clients connecting to the node using a 5GHz 4x4 160MHz ax wifi link with up to 1800/+ Mbps while having a powerline backhaul of up to 1000 Mbps?
A 5GHz 2x2 80MHz ax wifi has a link rate of up to 1200 Mbps which is better than every powerline.

The same consideration could be made for an ethernet 1 Gbps backhaul.

 

[OzarkEdge]

What should be the benefit of having clients connecting to the node using a 5GHz 4x4 160MHz ax wifi link with up to 1800/+ Mbps while having a powerline backhaul of up to 1000 Mbps?
A 5GHz 2x2 80MHz ax wifi has a link rate of up to 1200 Mbps which is better than every powerline.

The same consideration could be made for an ethernet 1 Gbps backhaul.

Gigabit Ethernet is full-duplex, 100% of the time, and not subject to interference.

WiFi is half-duplex, time shared, and subject to radio and object... even weather interference.

OE

 

[tgl]

Gigabit Ethernet is full-duplex, 100% of the time, and not subject to interference.

WiFi is half-duplex, time shared, and subject to radio and object... even weather interference.

Right. The rule of thumb that I've heard is that actual effective throughput of WiFi is not likely to be better than about two-thirds of the theoretical PHY rate. Wired ethernet comes a lot closer to reaching its full media speed. (Sadly, powerline seems to be more like WiFi here than like ethernet: IME the claimed PHY rates are not real-world results.)

But @kildare is right about the main point: there is no value in using the XT8s' 5GHz-2 radios for client service unless you've got sufficient bandwidth behind the AP to back that up. If you've got two high-bandwidth clients using both the 5GHz radios, you will just find that the bottleneck is now somewhere else, because you will have better than 1Gbps total throughput demand even after allowing for WiFi's inefficiency. I doubt there's going to be any value in setting up the RF that way and then using powerline for the backhaul. You'd need 2.5Gbps or better wired ethernet for backhaul, and after that you're just going to find that you're saturating your 1Gbps fiber service. Unless much of your bandwidth demand is local (pulling streaming media from your own NAS, say), I don't think it's going to make sense to try to use that second radio as client service rather than backhaul.

My advice is to just try to make sure that the high-bandwidth client devices connect to the primary node and not the satellite(s), so that they aren't subject to a backhaul penalty. This may or may not be feasible depending on your house layout, but it's worth thinking about. In the situation described, I'd be trying to use the satellite XT8 only to service low-bandwidth 2.4GHz devices.

 

[OzarkEdge]

Right. The rule of thumb that I've heard is that actual effective throughput of WiFi is not likely to be better than about two-thirds of the theoretical PHY rate. Wired ethernet comes a lot closer to reaching its full media speed. (Sadly, powerline seems to be more like WiFi here than like ethernet: IME the claimed PHY rates are not real-world results.)

But @kildare is right about the main point: there is no value in using the XT8s' 5GHz-2 radios for client service unless you've got sufficient bandwidth behind the AP to back that up. If you've got two high-bandwidth clients using both the 5GHz radios, you will just find that the bottleneck is now somewhere else, because you will have better than 1Gbps total throughput demand even after allowing for WiFi's inefficiency. I doubt there's going to be any value in setting up the RF that way and then using powerline for the backhaul. You'd need 2.5Gbps or better wired ethernet for backhaul, and after that you're just going to find that you're saturating your 1Gbps fiber service. Unless much of your bandwidth demand is local (pulling streaming media from your own NAS, say), I don't think it's going to make sense to try to use that second radio as client service rather than backhaul.

My advice is to just try to make sure that the high-bandwidth client devices connect to the primary node and not the satellite(s), so that they aren't subject to a backhaul penalty. This may or may not be feasible depending on your house layout, but it's worth thinking about. In the situation described, I'd be trying to use the satellite XT8 only to service low-bandwidth 2.4GHz devices.

There are theoretical considerations and there are practical considerations. If powerline is stable and fast enough for the application such that the consideration for using it is to avoid a niggling interference issue with using a theoretically faster wireless backhaul, then powerline could be the preferred solution.

Certain homework is left for the user!

OE

 

[RogerSC]

Not much good to say about powerline networking other than it usually will work, provides a connection. Speed is really questionable, though. Don't get enough speed on powerline networking here to make it worth my while. I am using it to connect the solar panel controller, where speed doesn't matter. Mesh backhaul use is not going to look good with powerline networking, wouldn't even cross my mind to go there.

Have you heard of MoCA? I'm using it here...not using it for our mesh backhaul because wireless backhaul works so well, but have tried it. Works great. MoCA uses the cable TV coax in your walls for hard-wired connections. Runs very close to ethernet cable speed, with a little added latency, couple of ms., from the MoCA adapters. Something that really works to think about *smile*.

 

[tbroome81]

Tried powerline and it slowed the speeds to the client devices down way too much. I'm just wanting to free up the faster AX 5-2 band for my AX devices instead of backhaul WITHOUT having to drill holes throughout my house to run CAT6 from one side of the house to the other. What about MoCA 2.5gbps adapters?

 

[Tech9]

I'm just wanting to free up the faster AX 5-2 band for my AX devices

How did you find your 5-2 band faster? Experiments?

 

[Tech Junky]

Tried powerline and it slowed the speeds to the client devices down way too much. I'm just wanting to free up the faster AX 5-2 band for my AX devices instead of backhaul WITHOUT having to drill holes throughout my house to run CAT6 from one side of the house to the other. What about MoCA 2.5gbps adapters?

How about buying a real AP instead of what you're using? I have a single AP covering 1300+ sq ft. The whole mesh / Wi-Fi backhaul is a joke as you can see from experience. Spending $1000+ on XT's is crazy and yet people fall for it all the time with the expectation of magic to occur when they plug them in.

With the AP I'm using I get consistent 1.5gbps in both directions on the LAN which is pretty good when you calculate the max of a 2400 link rate * 0.7 = 1.7gbps so, 200mbps isn't that much of a loss. The other 1/2 of the equation is the WIFI cards being used in your devices though. While my laptop hits 1.5gbps my phone might only hit 800mbps and something with an AC signal maxes out at ~700mbps. Some lower end devices though might only hit 100mbps.

 

[Tech9]

How about buying a real AP instead of what you're using?

AP connected how?

Ethernet backhaul isn't really an option given the construction of our house.

 

[tbroome81]

How about buying a real AP instead of what you're using? I have a single AP covering 1300+ sq ft. The whole mesh / Wi-Fi backhaul is a joke as you can see from experience. Spending $1000+ on XT's is crazy and yet people fall for it all the time with the expectation of magic to occur when they plug them in.

With the AP I'm using I get consistent 1.5gbps in both directions on the LAN which is pretty good when you calculate the max of a 2400 link rate * 0.7 = 1.7gbps so, 200mbps isn't that much of a loss. The other 1/2 of the equation is the WIFI cards being used in your devices though. While my laptop hits 1.5gbps my phone might only hit 800mbps and something with an AC signal maxes out at ~700mbps. Some lower end devices though might only hit 100mbps.

View attachment 44393

That'd be great, if I had ethernet connections in every room of my house, which is not the case. The only devices on my 5-1 band are AX devices. Everything not AX capable is on the 2.4 band.