Client Speed Test Upload over QoS Bandwidth Limit

[Seria17hri11er]

Hello. I am confused at the moment. After installing Merlin, QoS always seemed to limit a client's bandwidth during a speed test via browser etc. as it was set. I spent time testing and configuring before recently setting up spdMerlin and AutoBW. I could reproduce excellent results at Waveform bufferbloat test, and Cloud Flare's speed test.

Unfortunately after installing a lot of packages using AMTM, every time I test a desktop client [Cat5/6] the upload speeds are over the bandwidth limits resulting in poor bufferbloat/latency.

When I test via WifiMan 5 Ghz WiFi6 80 Mhz on my S21 it definitely seems to control things better.

I did have some of the packages I installed quit working, and I had to completely uninstall, remove data and reinstall to get working again.

I also did the install to add the "CakeQOS-Merlin" tab. Did I bork it doing that?

spdMerlin tests excluded from QoS, but results still look pretty darn good without the horrible latency I am getting when testing via browser at Waveform or Cloud Flare.

CenturyLink VDSL2 Fiber PTM "Quest Legacy"
Phoenix, Arizona, [East Valley]

Wan Packet Overhead: 30
MPU: 0
Mode: PTM
AutoBW

Download
Priority Q: besteffort
Flow Iso: Triple-iso
Nat Lookup: Yes

Upload
Same

Wash: No
Filter Dup Acks: No

Cake Statistics Look good, I think.

Download

​

Tin 0
Threshold Rate	95.57 Mbit/s
Target	5 ms
Interval	100 ms
Peak Delay	38 μs
Average Delay	3 μs
Sparse Delay	2 μs
Backlog	0 B
Packets Sent	194
Sent	68.39 KiB
Hash Indirect Hits	1
Hash Misses	56
Hash Collisions	0
Drops	0
ECN Marked Packets	0
Dropped ACK Packets	0
Sparse Flows	1
Bulk Flows	1
Unresponsive Flows	0
Max Packet Length	1,466
Flow Quantum	1,514

Upload

​

Tin 0
Threshold Rate	11.26 Mbit/s
Target	5 ms
Interval	100 ms
Peak Delay	2.31 ms
Average Delay	156 μs
Sparse Delay	3 μs
Backlog	0 B
Packets Sent	836
Sent	338.82 KiB
Hash Indirect Hits	18
Hash Misses	706
Hash Collisions	2
Drops	2
ECN Marked Packets	0
Dropped ACK Packets	0
Sparse Flows	207
Bulk Flows	133
Unresponsive Flows	0
Max Packet Length	1,392
Flow Quantum	343

Anyone got an idea of what is going on? Any tips would be greatly appreciated.

Recent speed tests:

Speedtest by Ookla - The Global Broadband Speed Test

Use Speedtest on all your devices with our free desktop and mobile apps.

www.speedtest.net

Speedtest by Ookla - The Global Broadband Speed Test

Use Speedtest on all your devices with our free desktop and mobile apps.

www.speedtest.net

Speedtest by Ookla - The Global Broadband Speed Test

Use Speedtest on all your devices with our free desktop and mobile apps.

www.speedtest.net

Speedtest by Ookla - The Global Broadband Speed Test

Use Speedtest on all your devices with our free desktop and mobile apps.

www.speedtest.net

With Gratitude

 

[L&LD]

Welcome to the forums @Seria17hri11er.

If your router/network was working as expected before adding additional scripts, does removing them bring back the expected performance?

If so, then don't add those scripts if they deteriorate your network experience.

 

[Tech9]

Anyone got an idea of what is going on?

No idea what's going on without router model and ISP plan details.

Unfortunately after installing a lot of packages using AMTM

This is how things are going downhill for most unexperienced users.

 

[Seria17hri11er]

Welcome to the forums @Seria17hri11er.

If your router/network was working as expected before adding additional scripts, does removing them bring back the expected performance?

If so, then don't add those scripts if they deteriorate your network experience.

Will test later. Thanks.

No idea what's going on without router model and ISP plan details.



This is how things are going downhill for most unexperienced users.

Great point! Lots of data points right.

GT-AX6000

ISP plan is for 100 MB down, I don't think they say crap about up.

 

[Tech9]

If you are confused already with your new router - run the stock firmware and get familiar with the features it offers first. It has more than enough features for average home user. If you need something else you don't find in stock Asuswrt - then look at Asuswrt-Merlin extra features. Jumping all in and with USB stick can produce undesirable results and makes your router as reliable as the USB stick.

About your ISP - 100Mbps down is okay, but "I don't think" is not usable piece of information. Perhaps your ISP is 100/10 and this is what you get from it. Online bufferbloat tests are mostly inaccurate and you can run in circles with QoS and scripts forever fixing something you perhaps don't need to fix for normal Internet user experience. Don't apply someone else's settings assuming you have the same issue.

 

[Seria17hri11er]

If you are confused already with your new router - run the stock firmware and get familiar with the features it offers first. It has more than enough features for average home user. If you need something else you don't find in stock Asuswrt - then look at Asuswrt-Merlin extra features. Jumping all in and with USB stick can produce undesirable results and makes your router as reliable as the USB stick.

Lol. I am confused about my test results, not my Router settings. lmao. Very aware.

About your ISP - 100Mbps down is okay, but "I don't think" is not usable piece of information. Perhaps your ISP is 100/10 and this is what you get from it.

CenturyLink makes no promises on the upload speed my dood. Your not getting my gist there?

As far as the VDSL connection:

Line 1:
61.440/9.727 Mbps

Line 2:
51.197/5.501

Testing, upload rarely sees 14, usually around 13. Never seen good latency results around that upload speed. Needs to be 11 or lower from my testing so far.

Online bufferbloat tests are mostly inaccurate and you can run in circles with QoS and scripts forever fixing something you perhaps don't need to fix for normal Internet user experience. Don't apply someone else's settings assuming you have the same issue.

That is what I seen only very few posts on, so I am happy to hear you confirm that. Basically sometimes you can get some good results with those test but too many variables right?

Welcome to the forums @Seria17hri11er.

If your router/network was working as expected before adding additional scripts, does removing them bring back the expected performance?

If so, then don't add those scripts if they deteriorate your network experience.

Thinking about this, I clearly didn't set up DNSCrypt for great performance and that wasn't helping.

 

[Seria17hri11er]

Don't apply someone else's settings assuming you have the same issue.

I haven't. Lot's of reading, trying to understand, testing and seeing results, or lack there of.

 

[Seria17hri11er]

Separate question for anyone.

Cake QoS. May I ask how you might set it up if you were me?

Wife and I work from home. I have a LG Thin Client for work, going into AWS plus a Cisco IP Phone. Wife is on a Dell Optiplex 9010 VPN into work which definitely consumes more bandwidth than my Thin Client. We both take phone calls.

3 Chromecasts on Wifi6 5g

Desktop Gaming PC

2 Cell Phones on Wifi6 5g

1 Chromebook on Wifi6 5g

1 RokuTV on Wifi6 2g



I've read about Cake settings numerous times, but hearing from others is always great.

Wan Packet Overhead: 30
MPU: 0
Mode: PTM

Download & Upload

Priority Queue: Best Effort
Nat Lookup: Yes
Wash: No
Filter Dup Acks: No
Custom: No
Enable Upload: No

Download:
Dual-dsthost

Upload: Dual-srchost

 

[Tech9]

Your not getting my gist there?

Quick search shows you can expect 100/10 depending on your distance from the nearest CenturyLink hub.

Cake QoS. May I ask how you might set it up if you were me?

No scripts, default Asuswrt-Merlin implementation, Manual settings 100 down and 10 up. Test and adjust if needed.

 

[Seria17hri11er]

Quick search shows you can expect 100/10 depending on your distance from the nearest CenturyLink hub.



No scripts, default Asuswrt-Merlin implementation, Manual settings 100 down and 10 up. Test and adjust if needed.

Sweet thanks. I was thinking to go back to it. I appreciate your assistance.

 

[Tech9]

Keep it simple and test one change at a time. If you don't have speed and latency issues with your normal Internet use - there is nothing to fix.

 

[Seria17hri11er]

Keep it simple and test one change at a time. If you don't have speed and latency issues with your normal Internet use - there is nothing to fix.

Completely correct. Then there are tweakers like me...who have to test everything lol

 

[Tech9]

Don't expect everything working the way you think or the way it was advertised.

The more you play with settings the faster you get to common reboot and reset experience.

 

[drinkingbird]

Separate question for anyone.

Cake QoS. May I ask how you might set it up if you were me?

Wife and I work from home. I have a LG Thin Client for work, going into AWS plus a Cisco IP Phone. Wife is on a Dell Optiplex 9010 VPN into work which definitely consumes more bandwidth than my Thin Client. We both take phone calls.

3 Chromecasts on Wifi6 5g

Desktop Gaming PC

2 Cell Phones on Wifi6 5g

1 Chromebook on Wifi6 5g

1 RokuTV on Wifi6 2g



I've read about Cake settings numerous times, but hearing from others is always great.

Wan Packet Overhead: 30
MPU: 0
Mode: PTM

Download & Upload

Priority Queue: Best Effort
Nat Lookup: Yes
Wash: No
Filter Dup Acks: No
Custom: No
Enable Upload: No

Download:
Dual-dsthost

Upload: Dual-srchost

The challenge your'e going to have is that if your speeds are not consistent, you either have to set it to the "worst case" and lose those times when you can get better performance, or set it higher and have it not help all the time. Comcast is a good example where they let you burst beyond your advertised speed for a certain number of megabytes then throttles back (mostly for speed test statistics to make them look good in the rankings). If you apply QOS at the advertised rate, you lose that benefit. Or in your case with DSL which fluctuates a lot, same thing.

For 100M download you're probably ok without QOS. 10M upload, depends on how much you're doing in the upstream direction. Video conferencing etc doesn't take as much as you might think Sometimes QOS can do more harm than good. But it is one of those things you'll just have to play with.

Don't be too concerned about bufferbloat scores. That's testing worst case scenario, and if your normal use doesn't match that, then the results are somewhat meaningless. If you regularly upload large files (or do online backups, lots of photos synching to google drive, etc) at the same time someone is video conferencing or using VOIP, then it is worth playing with, that's the main use case for home router QOS in my opinion.

I actually had QOS set up for my Mom's Ooma as when I was remotely fixing her PC and sending files back and forth the phone calls would get choppy (since I have to narrate to her exactly what I'm doing, lol). It may have made a slight difference, but not much. VOIP is very sensitive so the time it takes the QOS to do its thing (TCP has to detect the congestion and throttle back, but keeps trying to ramp up again) it just isn't enough to help much.

Now if you control both ends of the connection, QOS can be extremely powerful and helpful. In these cases where you're trying to control it all from one end, you lose a lot of the benefits.