WMM APSD Increases DL Transfer Speed

[amecca]

Hi all! I'm new this forum so please be forgiving

I just wanted to ask the community here for an explanation regarding some behavior I'm trying to understand with my system.

Network Configurations:
GT-AXE11000 With 2 ET8 Mesh Nodes
Merlin Was Installed Yesterday, Behavior Observed On Stock Firmware
2.4Ghz/5Ghz/6Ghz Separate Networks

I have been experimenting with my ul/dl speeds.

I'm finding that WMM APSD increases the speed on both my 5/6 Ghz networks by double. 600Mbps to 1200Mbps. From everything I'm reading, this feature is commonly explained to decrease network performance. Any idea why it would help so much?

In my scenario I'm performing a speed test from my zfold4 connected via 5Ghz and 6Ghz.

Just wondering your thoughts?

Let me know if you need any more details.

 

[OzarkEdge]

Hi all! I'm new this forum so please be forgiving

I just wanted to ask the community here for an explanation regarding some behavior I'm trying to understand with my system.

Network Configurations:
GT-AXE11000 With 2 ET8 Mesh Nodes
Merlin Was Installed Yesterday, Behavior Observed On Stock Firmware
2.4Ghz/5Ghz/6Ghz Separate Networks

I have been experimenting with my ul/dl speeds.

I'm finding that WMM APSD increases the speed on both my 5/6 Ghz networks by double. 600Mbps to 1200Mbps. From everything I'm reading, this feature is commonly explained to decrease network performance. Any idea why it would help so much?

In my scenario I'm performing a speed test from my zfold4 connected via 5Ghz and 6Ghz.

Just wondering your thoughts?

Let me know if you need any more details.

The x2 factor (2x600=1200) could suggest the difference is due to a x2 bandwidth change... if you fix the bandwidth to 80MHz and use a non-DFS control channel, does the speed stay steady over time? I'm ignoring the WMM APSD observation.

OE

 

[drinkingbird]

The x2 factor (2x600=1200) could suggest the difference is due to a x2 bandwidth change... if you fix the bandwidth to 80MHz and use a non-DFS control channel, does the speed stay steady over time? I'm ignoring the WMM APSD observation.

OE

Agreed, enabling the feature bounces the radios and restores the 160mhz bandwidth, but over time radar drops it back to 80. Probably will fall back down after a while of having APSD enabled.

 

[glens]

Also agreed. Maybe more like a 0.02 factor, certainly not 2.

 

[amecca]

Yeah, I agree with you guys.

I did some testing and it must be improving the bandwidth by temporarily setting 160Mhz. APSD does seem to be a red herring. I inevitably found the bandwidth dropped with APSD enabled as well.

I live in a VERY congested network area. It's been a nightmare. Honestly, I'm at the point where I think I may be out of options for getting over 500Mbps over wifi consistently. I've been staring at a channel survey and everything is pretty saturated.

Not to mention my ISP seems to have over-subscribed my neighborhood circuit. I pay for up to 2000Mbps DL and pull anywhere from 500-1000Mbps at the modem. (Usually closer to 600Mbps, looks like the 1200Mbps I got earlier was a fluke)

So between the ISP and dealing with the channel saturation, I'm lucky to get a third of what I can get.

Either way, thanks for all your help!

 

[drinkingbird]

Yeah, I agree with you guys.

I did some testing and it must be improving the bandwidth by temporarily setting 160Mhz. APSD does seem to be a red herring. I inevitably found the bandwidth dropped with APSD enabled as well.

I live in a VERY congested network area. It's been a nightmare. Honestly, I'm at the point where I think I may be out of options for getting over 500Mbps over wifi consistently. I've been staring at a channel survey and everything is pretty saturated.

Not to mention my ISP seems to have over-subscribed my neighborhood circuit. I pay for up to 2000Mbps DL and pull anywhere from 500-1000Mbps at the modem. (Usually closer to 600Mbps, looks like the 1200Mbps I got earlier was a fluke)

So between the ISP and dealing with the channel saturation, I'm lucky to get a third of what I can get.

Either way, thanks for all your help!

Since you're running merlin firmware there is an addon called "channelhog" and a few other similar scripts that when they detect the radio has dropped to 80mhz they bump it back to 160. However you might be able to get stable 160 by trying other channels (depending what channels are in your area, you only have one or two 160mhz to choose from).

In reality trying to maintain 160 is very difficult. However on a 2 stream AX client and 80mhz channel you should be able to get 800-900mbits in ideal conditions. But as you know lots of neighboring networks on the same channels will reduce that (however you have many more 80mhz channels to choose from, at least one of which does not overlap with DFS, hopefully 2, again depending where you live).

Also there have been reports that when no 160mhz clients are connected to the main router (and it drops to 80 as it normally does when none are connected), the aimesh nodes also drop to 80, even when you try to connect a 160mhz device, they stay at 80 because they just mirror the main router's behavior. That could be impacting you. You can try hardcoding the main router to 160, seems that works on some routers to keep it at 160 when no 160 device is connected, and doesn't help on others. Of course if radar is detected, it will always drop down no matter what you have set there.

As far as your ISP, yeah. With these gig and multigig tiers, especially on cable (or even fiber to the curb/cabinet) there is a lot of oversubscription going on. A call to the ISP may net you a solution, they might be able to add capacity or move you to another node, or they may just say "speeds are not guaranteed". In which case you can downgrade and save some money. Another poster in here recently got a refund and they are running new capacity to their area etc.

What do you get hardwired to the router? If you're still limited to less than what you get plugged into the ISP device, your router may just need a reset or some other settings checked.

You can move some devices to 2.4 (AX 2.4 is faster than N 2.4, but also more susceptible to interference) in order to take fuller advantage of your ISP bandwidth, and obviously wired devices will help too. With these high speed tiers, unless you're willing to move to AXe or BE, it isn't a matter of one client being able to saturate the connection, but your total clients. In fact, one client not being able to saturate it isn't necessarily a bad thing, it is a great form of QOS to make sure one device can't hog all the bandwidth.

 

[amecca]

Okay, I installed channelhog and checked my config. I tried setting to 160MHz for now. Everything is running okay for now around 750Mbps dl.

I'm currently based out of the U.S.

I'm currently running cable, soon, there is a new fiber provider installing in my part of town. When they are available, I'll see if I can give them a try. I may call my cable provider too, although not expecting them to be useful

I'll patch into my modem tomorrow and do another speed test and compare it to the GT-AXE11000 speed test. (using Ookla from the router interface) I did a config reset the other day but may try a full factory reset / wipe. (I didn't import my configs and did a manual reconfigure)

I have a bluetooth receiver connected to an Alexa Echo Link near one of my mesh nodes. I may try moving that as well to help with 2.5Ghz interference. (enabled bluetooth pre-emptive coexistence)

Thank you for your advice!

I'll respond back here with my findings.

 

[drinkingbird]

Okay, I installed channelhog and checked my config. I tried setting to 160MHz for now. Everything is running okay for now around 750Mbps dl.

I'm currently based out of the U.S.

I'm currently running cable, soon, there is a new fiber provider installing in my part of town. When they are available, I'll see if I can give them a try. I may call my cable provider too, although not expecting them to be useful

I'll patch into my modem tomorrow and do another speed test and compare it to the GT-AXE11000 speed test. (using Ookla from the router interface) I did a config reset the other day but may try a full factory reset / wipe. (I didn't import my configs and did a manual reconfigure)

I have a bluetooth receiver connected to an Alexa Echo Link near one of my mesh nodes. I may try moving that as well to help with 2.5Ghz interference. (enabled bluetooth pre-emptive coexistence)

Thank you for your advice!

I'll respond back here with my findings.

Better to use a PC for speed tests, the internal test can be problematic. On that router maybe not as much but still best to avoid it for the most accurate results.

Yes keep bluetooth as far away from your 2.4 wifi as possible, however since you're using 5ghz band for this, it shouldn't matter for this case.

 

[Tech9]

I've been staring at a channel survey and everything is pretty saturated

No, what you see is nearby APs tuned to specific channels. You don't see the bandwidth available per channel. This is a common mistake when selecting the best channel. It may be the one with multiple APs on it.

 

[glens]

Assume most other APs are running on "auto channel" and a desolate channel is so for a reason. Check it out, sure, but check the others as well.

 

[amecca]

Yeah, so I've been using the wifi analyzer app for Android.

It says that all the channels within 2.4Hgz and 5Ghz have 0/10 to 2/10 stars. Not sure if there is a better indicator I should be using?

Is there a specific method of determining channel saturation you'd recommend? Ive been trying to determine if things are okay by calculating SNR but new to this.

I was looking at the router interface and it says channel utilization 60-70 percent. Is this high?

Also, patched into the modem directly.... I'm getting around 200-300Mbps there. Makes no sense.... Still digging.

 

[L&LD]

Don't use an 'app' to find the best channel to use. They are all useless, particularly on a mobile device, as 99.999999% do not show channel utilization over time.

Don't try to overanalyze the 'numbers'. They are meaningless on uncalibrated equipment. And what is shown on the phone isn't what the router and/or client devices 'see' either.

Simply test each control channel, keep good notes, and choose the one with the lowest latency, followed by the best throughput and farthest reach.

 

[Tech9]

I was looking at the router interface and it says channel utilization 60-70 percent. Is this high?

It is high, but try 36-48 or 149-161 (if available) at 80MHz wide and watch what range usually shows less channel utilization. Wi-Fi spectrum is shared between all users and the utilization changes dynamically.