much higher buffer bloat on 2.4g compared to 5g

[jata]

Hey all,

This query is not specifically merlin 388.x related but I do need some advice.

I have an RT-AX86S now running merlin 388 with cake QOS enabled. My internet is pretty good - 80/40 FTTC and stable on speed tests that I run regularly so I have cake configured with correct up/down settings.

I have my work laptop (brand new with 160MHz capable wifi card) but it likes to switch between the 5g and 2.4g band.

When I do the waveform buffer bloat test, I get A or A+ on 5g but consistently low scores (C's) at the same location when it's connected to 2.4g

A also find similar situation with my macbook (also new) but I have to force it to use 2.4 for testing as it prefers 5g as I would expect.

So my question is why am I getting such bad buffer bloat on 2.4g vs 5g? How can I trouble shoot this? Should I even care?

 

[Tech9]

Should I even care?

No. The extra latency is coming from 2.4GHz band Wi-Fi. It's added between your router and the clients because the band is heavily used on limited available channels and your router has to wait for air time to transmit data. Test bufferbloat with wired devices only, but in general don't care much about it because it occurs at line saturation speeds only and most online testers are not very accurate. Not accurate on Wi-Fi - extra variable in the equation. Your Cake QoS is good enough, but test your Internet experience with and without it enabled. Cake QoS is incompatible with Runner/Flow Cache (NAT acceleration) and you may find your network more responsive without it. If you have no real issues without QoS - don't try to solve non-existing problems. There is a chance you are only hurting yourself 95% of the time because of something possibly happening 5% of the time.

 

[jata]

Thanks. Makes sense. I will turn off cake for a few weeks and see if I have any issues or complaints from the kids.

Where I live there is no congestion on either band. Maybe from my own aimesh nodes! So do think it is strange but I’m not going to lose any sleep over it.

 

[Tech9]

The best QoS is ISP upgrade - sufficient for your needs speed and FTTP connection. All router QoS options are in "hope for the best" category.

 

[Deleted member 84281]

Where I live there is no congestion on either band.

There is way more to congestion then just other WiFi users on the 2.4Ghz band.

 

[Tech9]

Yes, microwave ovens are usually on Channel 8. They leak enough RF to temporary render nearby 2.4GHz Wi-Fi useless.

 

[jata]

All good. Thanks guys. QoS is off and internet speed is 20% faster down and 15% faster up.

I guess the only time I might have an issue is when someone is downloading (taking all the bandwidth) while other family members are streaming video/facetime/zoom etc.

 

[drinkingbird]

All good. Thanks guys. QoS is off and internet speed is 20% faster down and 15% faster up.

I guess the only time I might have an issue is when someone is downloading (taking all the bandwidth) while other family members are streaming video/facetime/zoom etc.

You can use QOS. QOS will cause buffer bloat. That's the entire point. You are adding a forced restriction which will cause buffer bloat. It is doing exactly what it is supposed to do. If you need to re-enable it, ignore your buffer bloat, that figure is useless, your latency will always be higher when you saturate the line (which is what the test for buffer bloat does).

If your buffer bloat is A or A+ it means you are unable to saturate your line and something else is causing problems. And yes, latency will be higher on 2.4ghz, it has less bandwidth and more interference.

The only thing that will drive you more crazy than chasing speed test numbers is chasing buffer bloat scores.

 

[jata]

You can use QOS. QOS will cause buffer bloat. That's the entire point. You are adding a forced restriction which will cause buffer bloat. It is doing exactly what it is supposed to do. If you need to re-enable it, ignore your buffer bloat, that figure is useless, your latency will always be higher when you saturate the line (which is what the test for buffer bloat does).

If your buffer bloat is A or A+ it means you are unable to saturate your line and something else is causing problems. And yes, latency will be higher on 2.4ghz, it has less bandwidth and more interference.

The only thing that will drive you more crazy than chasing speed test numbers is chasing buffer bloat scores.

Haha. So true. I’ve had no issues with cake QOS and now I see how things go without any QOS.

I will stop testing in any case. Now I know it’s normal to get worse buffer bloat scores on 2.4g it is all good.

Thanks again everyone!

 

[drinkingbird]

Haha. So true. I’ve had no issues with cake QOS and now I see how things go without any QOS.

I will stop testing in any case. Now I know it’s normal to get worse buffer bloat scores on 2.4g it is all good.

Thanks again everyone!

Less QOS is always better but sometimes it is necessary. Enabling it only on upload if you have one of the like 1000/50 plans sometimes is needed, or enabling it on only one user who may be doing large downloads/uploads etc. But generally try to do either without it, or as little as possible. In reality, sometimes a physical restriction is a good way to deal with stuff. For example my outdoor AP that gets used by neighbors has a 100 meg connection back to my network, so they can only use 1/3 of my internet and I never have to worry about it impacting me. Not always possible or practical but something to keep in mind.

Professional equipment can do QOS quite well, especially when you have control over both ends of the connection. Shapers, RED/WRED, TOS, COS, all that good stuff has come a long way to try and make the restrictions as unnoticeable as possible. But obviously when you do hit a restriction, whatever traffic has been given the lowest priority will suffer.