Getting F Grade on Waveform / Bufferbloat Test?

[thisisreallife]

Under Connection Benchmark I'm getting 3 A+s. However under Waveform I'm getting F Grade, how can I fix this?

A few weeks ago it was working fine and I was getting a grade of A+.

The only noticeable thing I've done it set up my Access Points. That shouldn't be causing this issue right? 

Connection Benchmark Results

Bufferbloat Waveform Results

Notice my Upload is really high, +600ms.

How can I fix this issue?

EDIT:

Ok I've just done another test, I turned off Congestion Control (put it on Never) and did the test and I got a test of A on Bufferbloat/Waveform...

It seems like QoS 'Upload' is causing some issue somewhere. If I put Upload on 100% and keep Download on 50% like it's telling me, I get a score of A on Waveform.

However once I turn Upload down to 99%, the Upload Active on the test jumps to +600ms and I get a rating of F.

I'm using this version: R2 3.3.280. Should I upgrade to a beta version instead? 

[DARKNESS]

Yeah beta version is better currently since it prioritizes traffic more efficiently do to the DPI though I'd go with version 3.3.502 since it more stable though if your able to run the v3.3.615 then stick with it though it known for taking long to access interference initially

[Netduma Fraser]

With QoS enabled bufferbloat can't happen so that 600ms upload would never happen. The QoS on the router can mess with 3rd party tests like that so best to use Benchmark or do a manual test using Wireshark to optimize QoS as much as possible. 

[thisisreallife]

17 minutes ago, Netduma Fraser said:

With QoS enabled bufferbloat can't happen so that 600ms upload would never happen. The QoS on the router can mess with 3rd party tests like that so best to use Benchmark or do a manual test using Wireshark to optimize QoS as much as possible. 

Thanks!

Do you recommend I upgrade to one of the betas also?

Also I'm sure I read that on the beta you can only use the mobile app version and not the normal interface. Is this true?

[DARKNESS]

@thisisreallifeYou can access it through a mobile browser or on desktop by using http://192.168.77.1/dumaweb/

 

[Netduma Fraser]

27 minutes ago, thisisreallife said:

Thanks!

Do you recommend I upgrade to one of the betas also?

Also I'm sure I read that on the beta you can only use the mobile app version and not the normal interface. Is this true?

It definitely couldn't hurt, we did make some changes in the beta that should allow it to get better results on those 3rd party tests.

You can only use the QoS on the mobile view for the moment but everything else is available through the desktop view.

[thisisreallife]

53 minutes ago, Netduma Fraser said:

It definitely couldn't hurt, we did make some changes in the beta that should allow it to get better results on those 3rd party tests.

You can only use the QoS on the mobile view for the moment but everything else is available through the desktop view.

Oh so I log into the mobile view to edit QoS and then log into the desktop for all else?

I can essentially have two tabs open then.. One that's mobile and one that's desktop and it will work fine?

Which is the most stable beta version to download? version 3.3.502 as Darkness said?

[Netduma Fraser]

48 minutes ago, thisisreallife said:

Oh so I log into the mobile view to edit QoS and then log into the desktop for all else?

I can essentially have two tabs open then.. One that's mobile and one that's desktop and it will work fine?

Which is the most stable beta version to download? version 3.3.502 as Darkness said?

You can also use some of the other features on the mobile view but yes you could do that.

I'd generally recommend the latest which is .615

[AZX]

Have you ever considered how the congestion control algorithm for TCP in Windows or a NIC affects the results of tests? I can't recall seeing any mention of disabling it or any discussion about optimizing network settings for use with connection benchmark or any other bufferbloat test. Most users, including myself, tend to perform these tests on Windows, and the configuration of the operating system's network settings is important to factor.

For instance, by default, Windows uses the Cubic congestion algorithm, which is acceptable. However, I've observed significantly better results in bufferbloat testing using tools like Waveform, Devina, and Connection Benchmark with CTCP or even without any congestion control mechanism. More recently, I've been experimenting with BBR2, which is not widely available (available in Windows 11 version 22H2 or later), but the outcomes are consistently and substantially better in terms of both throughput and latency compared to the default Cubic algorithm in Windows.

Certainly, there are other settings within Windows that can influence bufferbloat test results. Nonetheless, the switch from Cubic to BBR2 has yielded the most notable improvements in my experience. With all this said, I'm still uncertain about the specific network configuration that Netduma suggests or recommends for use in Windows, especially considering the inclusion of features like QoS and ping optimizer in DumaOS.

 Can you give some insight @Netduma Fraser? 

 

Also this is how to implement BBRv2 in Windows 11 if anyone wants to test.

-Open CMD/Powershell as admin.

-Copy and paste one line at a time.

netsh int tcp set supplemental Template=Internet CongestionProvider=bbr2
netsh int tcp set supplemental Template=Datacenter CongestionProvider=bbr2
netsh int tcp set supplemental Template=Compat CongestionProvider=bbr2
netsh int tcp set supplemental Template=DatacenterCustom CongestionProvider=bbr2
netsh int tcp set supplemental Template=InternetCustom CongestionProvider=bbr2

-(Powershell only) display current/updated congestion algorithm.

Get-NetTCPSetting | Select SettingName, CongestionProvider

[Netduma Fraser]

@AZX To my knowledge we have never tested the QoS offerings on Windows and their effect. I'm not sure it would be worth it to be honest, QoS on the router works on the network level for all of your devices, whereas the Windows QoS of course only works for that singular device, as long as QoS is optimized for the entire network for us that is the main thing. It may be worth it for people to experiment with but as long as QoS on the router is enabled bufferbloat won't occur.

[AZX]

@Netduma Fraser Are you saying you don’t believe your team has tested whether Windows conflicts with dumaOS Connection Benchmark? My concern is that testing buffer bloat in Windows could yield varied results due to the different implementations of QoS in Windows. Couldn’t it lead to results that do not accurately reflect the relative performance of the router and its QoS implementation?