1.5Gb fiber connection with 3 mesh nodes. Clients getting ~500Mbps. Setting or nature of AIMesh?

[neil0311]

3 node AIMesh RT-AX88U Pro using wired backhaul with 2.5Gb ports through 2.5Gb switch (link status shows 2.5Gb). FTTH connection is 1.5Gbps. Using Smart Connect and the wireless log shows on the router shows 160M connections on 5GHz with 2Gb or better port speeds, yet speed tests through the ISP show that clients seem to be limited to roughly 1/3 speed (~500Mbps).

Wired connections to any of the 3 routers/nodes shows essentially the same thing. A speed test directly from the router shows 1.6Gbps up/down so the fiber is provisioned correctly to the ONT.

Is there a setting I need to tweak? Is this expected behavior on a mesh setup?

 

[TheLostSwede]

Have you tried running iPerf3 on the local network to see if you can pin-pint any local issues?
You need two computers to install it on, run one as a server and one as a client.
Easiest way to figure out if there's a local problem or not.

External speed tests aren't really going to help you figure out internal problems.

Also, if all your APs are wired, why bother with AIMesh? Just set them up as APs, your wireless devices will still roam between them.
Been running various bits of hardware as APs for years, all wired backhaul due to poor signal between floors where we use to live.

 

[bbunge]

Can you eliminate the switch and connect Ethernet to the nodes directly to the router? Do you have Ethernet backhaul mode enabled?

 

[neil0311]

Have you tried running iPerf3 on the local network to see if you can pin-pint any local issues?
You need two computers to install it on, run one as a server and one as a client.
Easiest way to figure out if there's a local problem or not.

External speed tests aren't really going to help you figure out internal problems.

Also, if all your APs are wired, why bother with AIMesh? Just set them up as APs, your wireless devices will still roam between them.
Been running various bits of hardware as APs for years, all wired backhaul due to poor signal between floors where we use to live.

Interesting thought. I don’t think the issue is LAN related but I can check.

So what does AI Mesh give me that running them as pure APs wouldn’t. One thing I can think of wireless backup for backhaul, if the switch fails, but that’s a tiny risk.

I do like the consolidated logging and client list. But none of that is more important than performance.

 

[TheLostSwede]

Interesting thought. I don’t think the issue is LAN related but I can check.

I just figured it was an easy way for you to check if there's something wrong internally.

So what does AI Mesh give me that running them as pure APs wouldn’t. One thing I can think of wireless backup for backhaul, if the switch fails, but that’s a tiny risk.

I do like the consolidated logging and client list. But none of that is more important than performance.

Not sure to be honest, I have never used AiMesh, as I haven't had any need for it. The client list and logging seems to be one benefit, but the failover sounds like something you're unlikely to ever end up benefiting from, unless someone damages a cable or unplugs something from the network.

However, as mentioned, roaming works just fine without it, as long as the network has the same name and channel everywhere. Well, in theory you don't even need the same channel, but it tends to be smoother if you use the same channel in my experience. I've had Netgear, TP-Link and Asus gear all working together just fine. That said, I have had separate 2.4 and 5 GHz networks, so if you use a single network name for all frequencies, AiMesh might help.

It's at least worth a try if you can't work out what's wrong, maybe try with one AP first and see if it makes any difference?

 

[neil0311]

Can you eliminate the switch and connect Ethernet to the nodes directly to the router? Do you have Ethernet backhaul mode enabled?

I don’t have Ethernet backhaul enabled. Figured having wireless backhaul as a failover would be a good idea but can’t try it.

Using the switch is the only way to get 2.5Gb LAN speeds between the devices. Otherwise one of the devices would be limited to 1Gb which at this point I guess they seem to be anyway.

 

[OzarkEdge]

I don’t have Ethernet backhaul enabled. Figured having wireless backhaul as a failover would be a good idea but can’t try it.

Try disabling wireless backhauls (Ethernet Backhaul Mode enabled) and test again.

OE

 

[neil0311]

Try disabling wireless backhauls (Ethernet Backhaul Mode enabled) and test again.

OE

No change. Tried “auto” as well and same behavior. Both wired and wireless clients get about 500Mbps up/down. QoS is off.

WTF? My bandwidth has increased and my clients are getting less throughput.

 

[neil0311]

I just figured it was an easy way for you to check if there's something wrong internally.

Not sure to be honest, I have never used AiMesh, as I haven't had any need for it. The client list and logging seems to be one benefit, but the failover sounds like something you're unlikely to ever end up benefiting from, unless someone damages a cable or unplugs something from the network.

However, as mentioned, roaming works just fine without it, as long as the network has the same name and channel everywhere. Well, in theory you don't even need the same channel, but it tends to be smoother if you use the same channel in my experience. I've had Netgear, TP-Link and Asus gear all working together just fine. That said, I have had separate 2.4 and 5 GHz networks, so if you use a single network name for all frequencies, AiMesh might help.

It's at least worth a try if you can't work out what's wrong, maybe try with one AP first and see if it makes any difference?

I used to use 2 separate networks for 2.4 and 5GHz, but decided to try one SSID and smart connect because of the mesh and IOT devices.

I had better client perf with my Comcast connection and one router.

 

[jerry6]

if you had better perf using one comcast router why not getrid of mesh and "smart" connect assign clirents a channel each in2.4 and 5 ghz

 

[neil0311]

if you had better perf using one comcast router why not getrid of mesh and "smart" connect assign clirents a channel each in2.4 and 5 ghz

The Comcast scenario was at my old house. We just closed on a new home with 3 stories and 5600 sq ft.

Doing what you said is an option but I wanted to keep 2.4 for IOT and other devices that don’t need high speed. I also like the Smart Connect roaming.

My iPhone now is maxing out the 80 MHz wifi 6 channel at close to 600 Mbps. It’s the wired clients that are puzzling to me. Wired Gb should get ~930Mbps with no other activity and a 1.5Gb pipe to the ISP.

 

[sfx2000]

If you're on WiFi - that's basically what you get. 500 Mbit/Sec isn't bad...

Look at wired performance and go from there...

 

[jerry6]

The Comcast scenario was at my old house. We just closed on a new home with 3 stories and 5600 sq ft.

Doing what you said is an option but I wanted to keep 2.4 for IOT and other devices that don’t need high speed. I also like the Smart Connect roaming.

My iPhone now is maxing out the 80 MHz wifi 6 channel at close to 600 Mbps. It’s the wired clients that are puzzling to me. Wired Gb should get ~930Mbps with no other activity and a 1.5Gb pipe to the ISP.

what is the house made of concrete ? my house in the city is 6000 sq feet concrete basement brick then cemet blocks and wood frame and I get full coverage using one router , I like to keep things as simple as possible . wired should get better perf for sure I'm getting 800 with AX clients close by the router. 300 to 500 or more with AC clients depending on the floor and how many walls / doors between router and clients . 1000/100 connection . good luck

 

[neil0311]

If you're on WiFi - that's basically what you get. 500 Mbit/Sec isn't bad...

Look at wired performance and go from there...

Not to be rude, but did you read any of my posts in the thread?

 

[neil0311]

what is the house made of concrete ? my house in the city is 6000 sq feet concrete basement brick then cemet blocks and wood frame and I get full coverage using one router , I like to keep things as simple as possible . wired should get better perf for sure I'm getting 800 with AX clients close by the router. 300 to 500 or more with AC clients depending on the floor and how many walls / doors between router and clients . 1000/100 connection . good luck

Walkout basement on partial concrete foundation on slab. Standard stick framing.

There’s no way the router in the basement which connects to the ONT could service the 2nd floor. I have the 2nd floor mesh node connected in my media room, via wired back haul to the basement router, and my TV, A/V receiver, Blu Ray, and DVR connect via Ethernet cable to the mesh node sitting there as well.

The questionable mesh node is the one of the first floor. Might not be needed and may try turning it off and see how it goes.

 

[neil0311]

Weird, so now I'm getting more upload than download. At the main router, the testing still shows full provisioned speed both up and down.

 

[Ripshod]

Weird, so now I'm getting more upload than download. At the main router, the testing still shows full provisioned speed both up and down.

View attachment 58168

That first speedtest. Is that through the mesh node connected through the switch? Something is introducing massive amounts of latency to your internal network. If it isn't the switch (test with a back-to-back addapter),then it looks to me like you have at least one parallel path (DHCP storm anyone?).
Also check you don't have any 568A to 568B crossover cables in the mix. If you've installed the cables yourself re-check the terminations.

 

[neil0311]

That first speedtest. Is that through the mesh node connected through the switch? Something is introducing massive amounts of latency to your internal network. If it isn't the switch (test with a back-to-back addapter),then it looks to me like you have at least one parallel path (DHCP storm anyone?).
Also check you don't have any 568A to 568B crossover cables in the mix. If you've installed the cables yourself re-check the terminations.

The first test is a client using wifi connected to one of the mesh nodes connected via the 2.5Gb WAN port, through a 2.5Gb capable switch.

The wiring in the house was run by an electrician and some of the jacks were punched down by the fiber install tech.

No crossover cables used, and all patch cables are Cat 6.

I guess one test I can run is to remove the mesh nodes (could I just power them down) and test directly from the router in the basement. Currently, wired clients connected to that router still get throttled for some reason.

 

[newb]

3 node AIMesh RT-AX88U Pro using wired backhaul with 2.5Gb ports through 2.5Gb switch (link status shows 2.5Gb). FTTH connection is 1.5Gbps. Using Smart Connect and the wireless log shows on the router shows 160M connections on 5GHz with 2Gb or better port speeds, yet speed tests through the ISP show that clients seem to be limited to roughly 1/3 speed (~500Mbps).

Wired connections to any of the 3 routers/nodes shows essentially the same thing. A speed test directly from the router shows 1.6Gbps up/down so the fiber is provisioned correctly to the ONT.

Is there a setting I need to tweak? Is this expected behavior on a mesh setup?

 

[newb]

how do you do 2.5G connection on RT-AX88U? Please a schematic pic would be helpful. Thanks.