RT-AX86U 2.5G port not giving close to 2.5G speed - troubleshoot

[clownfishman]

Just purchased the RT-AX86U and using ASUS firmware 3.0.0.4.388_22525-gd35b8fe. With adaptive QoS turned off (turned on will only get 1.5G/950)

SETUP:
[Bell Fibe 3.0 Gbps, with 10G port] ---- Cat 6e, (tested with usb 2.5G NIC, got ~2.5G) ---- RT-AX86U (with 2.5G WAN port).

Using the ASUS admin console, speedtest on the router only able to get max 1.6G/2.2G max - so wondering if there is anything I can tune to make it full/close to 2.5G? (Tested the line, it is capable of close to 2.5G down/up; upstream Bell router is also at 3.1Gbps/3.1Gbps)

Speedtest from RT-AX86U samples (down/up):
1616/2154
1062/2197
1426/2131
1178/2164

 

[Adooni]

router speed test is more for checking connection quality/ no packet lost etc. It is hard to get real results as it demands of router CPU load etc.

to check what is your ISP real speed you should use web based pages like for example speedtest.net , https://www.waveform.com/tools/bufferbloat etc.

 

[clownfishman]

router speed test is more for checking connection quality/ no packet lost etc. It is hard to get real results as it demands of router CPU load etc.

to check what is your ISP real speed you should use web based pages like for example speedtest.net , https://www.waveform.com/tools/bufferbloat etc.

so with RT-AX86U router, it only has a 2.5G WAN port and no LAN is capable of 2.5G, so max testing from any 1G LAN is 9xx/9xx with webpage.

What I need to know is why from router it is not getting full 2.5G (the upload seems to be able to reach it without issues and there is no extensive download happening). So unless there is a way to do a speedtest from router to internet, cannot determine the real speed

 

[RMerlin]

Router's CPU probably not able to keep up.

 

[clownfishman]

Router's CPU probably not able to keep up.

That is unlikely also, it have no problem showing the ~2.1 Gbps speed on the upload speed, so seems like something is hogging the download speed.

The slight improvement comes from shutting down:
AiProtection
QoS & GeForce Now
Web History
Bandwidth Monitor

But still, download only go to average 1.8G and can't go higher, so something else must be limiting

 

[gp-se]

Bell Canada forum | DSLReports, ISP Information
A lot of others experiencing the same issue.

 

[clownfishman]

Bell Canada forum | DSLReports, ISP Information
A lot of others experiencing the same issue.

Maybe you have misunderstood the issue - I do not see how it is a Bell issue.

(1) Bell HH4000 modem tested comes back at [3.1Gbps/3.1 Gbps]
(2) Bell HH400 modem with 10G port >>>> 2.5G USB NIC card to a laptop [~2.5 / ~2.5] (that means the cable and upstream is also giving max speed

Only when (3) Bell HH4000 10G port >>> Asus RT-AX86U on 2.5G port, I am getting [1.8/2.1] on modem's speedtest.net , so it has to the config of the asus that is not supporting the max speed of that line, so wondering what config could have slow it down.

 

[gp-se]

Maybe you have misunderstood the issue - I do not see how it is a Bell issue.

(1) Bell HH4000 modem tested comes back at [3.1Gbps/3.1 Gbps]
(2) Bell HH400 modem with 10G port >>>> 2.5G USB NIC card to a laptop [~2.5 / ~2.5] (that means the cable and upstream is also giving max speed

Only when (3) Bell HH4000 10G port >>> Asus RT-AX86U on 2.5G port, I am getting [1.8/2.1] on modem's speedtest.net , so it has to the config of the asus that is not supporting the max speed of that line, so wondering what config could have slow it down.

[Internet] Have Bell 3Gbps internet but only getting 2.4Gbps - Bell Canada | DSLReports Forums
GigaHub drops download speed only - Bell Canada | DSLReports Forums
Other Bell users have reported similar speed issues when using a third party router.

 

[tgl]

That is unlikely also, it have no problem showing the ~2.1 Gbps speed on the upload speed, so seems like something is hogging the download speed.

Keep in mind that @RMerlin generally knows what he's talking about in this realm . I have no problem buying his explanation, because there isn't necessarily the same CPU load involved in download and upload. For example, unless you're running with no firewall at all (not a good idea) there will most likely be more packet-filtering logic active on the download side.

What I'd try given your port configuration is concurrent internet speedtest runs from several clients plugged into the LAN ports, and maybe some wireless clients too. The reason you bought this hardware was obviously not to get >1Gbps throughput to any one client, it was to be able to sustain high rates to several clients at once --- so test it that way, and see what happens.

BTW, I have no experience with the RT-AX86U, but I previously found out that on my XT8 units, the total bandwidth available from the 2.5G WAN port to the three 1G LAN ports is only about 1Gbps. The higher-bandwidth WAN port is not entirely a waste of money, because there's additional bandwidth available to the wireless hardware. You might find out that ASUS skimped similarly on the RT-AX86U's switching fabric, or you might not; test it!

 

[scooter2765]

I stopped using the 2.5GB port in mine and now wish I had not paid for it. When I use it on mine in AP mode, I only use AP mode. Wireless clients very often stop being able to connect to anything but the router web page itself. I also see the network map lose all it's clients in the count. Then I have to reboot the router to get stuff back. I switched to a 1GB port for the uplink and this stopped happening. Also I saw no speed improvement using the 2.5 GB port at all. Seems to me something there is half baked.

 

[tgl]

... I switched to a 1GB port for the uplink and this stopped happening.

Interesting. What equipment is on the other end of your uplink cable, exactly?

Probably unrelated, but: I've been hitting 2.5G compatibility problems with a new Mac Mini with a 10G ethernet port. It works pretty well when plugged into a 10G switch port (success with two different switches, here) but when plugged into a TRENDnet 2.5G switch it is seriously flaky --- every now and then it just loses connection for ten to thirty seconds. Hard to tell whose fault that is. Now I wonder if the 2.5G ethernet spec is less well nailed down than the other speeds.

Also I saw no speed improvement using the 2.5 GB port at all. Seems to me something there is half baked.

How were you testing? It's pretty easy to believe that the uplink cable wouldn't be the bottleneck in many situations.

 

[scooter2765]

the upstream switch is a QNAP QSW-m2108-2c . It is linked to my core 10g Microtik CRS309-1G-8SFP+ with fiber and that is linked 10GB twinax to the untangle router which is linked 2.5GB to a motorola MB8611 cable modem. I don't have any other 2.5gb ports to plug it into either I do have a couple SFP+ to 10GBE SFP's but they definitely are flaky at 2.5GBE speeds so I don't trust them for testing. The Qnap with the latest firmware works really well with all the other 2.5gb devices I have it conected to ...all two of them one's a intel nic the other is a moca adapter..they both work well.

I didn't do any real good testing. I have an internal openspeedtest docker container on a proxmox server that is 10GB to the core I use for quick speed tests using it with my wifi 6 phone I don't even get 9xx speeds with the ASUS. I get about 780mbs...I got the same with the 2.5 gb port when it worked but it dropped and locked too much. It would drop every time I tried to go from my phone to the mangement ip of the qnap switch. It has to be some strange network bug with the 2.5gb port. Since I only use it for wifi and I get near gig speeds I was not inclined to troubleshoot further but something is definitely not kosher with that port or the config of it.

 

[scooter2765]

Are you guys getting wifi speeds over a gig at all with the ax86U or are your speed tests aggregates of multiple 1GB ports? Like running multiple iperf3 from hosts.