CAKEQOS/FLEXQOS Advise/Guidance/Best Practice

[marcd5123]

Hi everyone apologies if this may or may not have been mentioned on previous threads/posts I have actually tried to read lots of posts but could not quite get the answer I was looking for.

I have a Asus RT-AX86U Pro and running latest MerlinWRT..I have a EE broadband FTTC ADSL2 Connection of 80/20 but get a solid/ consistent speed tests of 74DL and 18.5UL .

So do know and understand about CakeQos and FlexQos they are very indeed very different to how they both run as a "Qos" .

I had been running CakeQos for a good few weeks and yes did get some decent response from it and managed to get my Bufferfloat down to a grade of A+ but at heart I understand Cake is really more of a proper "fairness" control.

So I decided to learn and teach my self about adding on to my router and try out FlexQos just to see how different the results would be .

Now this considering I mainly use my bandwidth for 4K streaming Netflix,Prime ,Apple TV ...but alot of using Kodi with Diggz Build and RD account with my Nvidia Shield Pro 2019 unit .

I understand that FlexQos works along side or uses parts of Asus adaptive Qos in order to works and FlexQos definitely compared to Cake Is way way more powerful and sophisticated program were like others you can tinker or tweeks till the cows come home .

I do understand that Cake disables all Hardware Acceleration so to my knowledge and understanding this actually uses far more system resources like the CPU which could be good or bad depending how you look at it I suppose.

One of the reasons I wanted to give FlexQos a try is to see just how well the "adaptive" side of flex works as it's supposed to be far better at using system resources so perhaps less stress on the router. Yes I understand it does use hardware acceleration but as it's a "adaptive" Qos this means it should be adapting and using system resources only when needed to a device that needs it so its far more efficient as a Qos in principal ???

Cake obviously gets the router to constantly use resources to keep every client connected with equal amounts of bandwidth available as it's a fairness control. Unless I'm mistaken then please do correct me so presents more stress to the router and resources .Yes for me in principal cake would be the better option but I'm not so sure that's the case .

I've been running FlexQos for almost 1 weeks now and it's been very good experience so far and I way go as far as stating that my Kodi Streaming experience has been noticeably better especially when buffering is concerned for 4K streaming.

I have attached a picture of the settings I have used so far but I have not changed anything in the device table as of yet as that's just on default settings. Just to see how well FlexQos performs without hardly any modifications done to it.

I used the custom control and made one or 2 adjustments that all .

When I run FlexQos now I get a ratting of C so their is higher latency then when I had cake running as that was A+ .

With Flex is set my DL as 74 and UL as 18 as this is what I got after running speedtest from the router and my mobile device as a consistent result .

Oddly enough even though the latency is still higher than CakeQos it's still miles lower than when I run it without any Qos enabled so I'm still benefiting with using Cake is this a slight tradeoff ???.

So should I change that part of I suppose a bit more latency does impact Flex that much as it's adaptive.

Does this look correct is this wrong or should I try something else .

Many thanks all

 

[Tech9]

Don't look at this Waveform website and better run Cake QoS. FlexQoS is Adaptive QoS based and depends on Trend Micro signatures file. Many users find it broken from time to time and the developer himself recommends CakeQoS for WAN speeds under 300Mbps.

 

[marcd5123]

Don't look at this Waveform website and better run Cake QoS. FlexQoS is Adaptive QoS based and depends on Trend Micro signatures file. Many users find it broken from time to time and the developer himself recommends CakeQoS for WAN speeds under 300Mbps.

Hi thanks for the reply. I'll take that onboard as part of my research as their are so many mixed opinions . I'm looking forward to seeing what other also have to say as well .

So far for me 1 week I'm getting some interesting results putting aside "Waveform" website I'm not that fussed about that part .But flexqos at the moment seems to be handling my Kodi Diggz Build streams significantly better than cake ...so it's definitely subjective it's early days for me . I had cake on for a few weeks so to give FlexQos the benefit of the doubt and a fair test period I will do the same .

 

[Tech9]

It may be working well for you at the moment and get broken with the next signature file update. This is the problem. Asuswrt-Merlin Cake QoS option is simple and more consistent. No matter what research you do and how long you play with QoS types and settings the best QoS is changing your ISP line to symmetrical fiber, if available. You can't improve much the quality of an ISP line just manipulating one side of the connection.

 

[Crimliar]

This is going to be working from memory as I no longer need to do this myself!
The best way to set up parameters for CAKE QOS, is to also install spdMerlin. You can then auto-run spdMerlin on a custom schedule at times when there is no one home/using the internet. My schedule just used to run 4 or 5 times a night. You can then have the router through CAKE take an average and set it's limits to 90-95% of that average. That used to work very well for me on Vodafone VDSL2 9(you are on VDSL2 not ADSL2), but it requires that you have a stable connection. If you are on an oversubscribed street cabinet, you may need to drop the threshold 5% at a time until it works.
*What modem are you using?

 

[Shadow101]

Im just gonna bump this thread as I see no need flooding the forum , I got a weird scenario , I pay for a 200/200 symetrical line but my isp one day after checking my fibre lines upgraded me? to a 500 down and 100 up late last year , the download speeds can vary from going anywhere between 500-800 reportedly , now hate me for this but I did use the waveform bufferbloat test and afew others and I get a B rating consistently , my download latency is prefect when testing on speedtest sites with latency between 10-15ms but upload at 100mbps I get anywhere between 50-79ms which is quite high , so I tried flexQos for the first time in years and found if I lowered my upload by just 5mbps I get a low latency of 10-20ms , great in this context but now I have to lower the download speed as latency doubles , what would be my best in terms of keeping the upload unsaturated and the download the same? it doesn't completely bother me but would be nice having a little more stable voip and video calling but keep my high speed downloads for games etc

 

[Crimliar]

It's quite probable that the only time you are saturating your connection is when you are performing speedtests or bufferbloat tests, and even then only from relatively new hardware. All too often network performance tests will be stressing the hardware used to make the test and not your connection.

 

[Shadow101]

It's quite probable that the only time you are saturating your connection is when you are performing speedtests or bufferbloat tests, and even then only from relatively new hardware. All too often network performance tests will be stressing the hardware used to make the test and not your connection.

mmm that may be so maybe Im just overthinking a problem I don't really have , heres a before and after though
Before :
https://www.waveform.com/tools/bufferbloat?test-id=e0cde156-fc5f-447d-8e00-96845846c4c9 QOS disabled and flexQos not installed yet
After:
https://www.waveform.com/tools/bufferbloat?test-id=d0d2c626-5fd3-46e3-bcd5-ce40a8c2e2c5 reduced upload to 75mbps

I think i will just uninstall Flex then , I also see Dave archived the github for flex might he be done with the project as a whole?

 

[Jordan123]

I'm aware every connection has different variables but can I get a screenshot of the settings you used for Cake? I'm with BT with an 80/20 rough split and I'm very lucky to get an "A" rating?

Usually falls at a B rating sadly, thanks in advance.

 

[Tech9]

I'm very lucky to get an "A" rating?

Do you have any real connection issues or just chasing the Waveform score?

 

[marcd5123]

This is going to be working from memory as I no longer need to do this myself!
The best way to set up parameters for CAKE QOS, is to also install spdMerlin. You can then auto-run spdMerlin on a custom schedule at times when there is no one home/using the internet. My schedule just used to run 4 or 5 times a night. You can then have the router through CAKE take an average and set it's limits to 90-95% of that average. That used to work very well for me on Vodafone VDSL2 9(you are on VDSL2 not ADSL2), but it requires that you have a stable connection. If you are on an oversubscribed street cabinet, you may need to drop the threshold 5% at a time until it works.
*What modem are you using?

Well spotted yes typo !! I meant VDSL2 ..thanks for that information..

 

[Jordan123]

Do you have any real connection issues or just chasing the Waveform score?

Given the options you've given me; Score. but kind of falls into a pride thing too because I've noticed multiple posts with pretty much the same connection as mine getting A+ whilst I've wasted so much time on it with the best score being A and if I can rule out the issue being my fault I'll be happy to move on with my life

 

[Tech9]

I've wasted so much time on it with the best score being A

You've wasted time and perhaps hurt yourself in the process. This site has to be blocked by AiProtection as malicious. Many people are chasing ghosts because of it with some not only wasting time, but also money in unnecessary hardware upgrades.

 

[Jordan123]

You've wasted time and perhaps hurt yourself in the process. This site has to be blocked by AiProtection as malicious. Many people are chasing ghosts because of it with some not only wasting time, but also money in unnecessary hardware upgrades.

That seems fair; Real usage it's performing well and I'm loving not being bombarded with advertising now due to AdGuard but like you said seems I'm chasing ghosts because for the life of me I'm never getting that A+ no matter how I've tweaked it.

Seems counterproductive but I'll hold on to see if the settings are shared by OP and if I get B results again I can finally stop beating a dead horse as they say

 

[dave14305]

Yes for me in principal cake would be the better option but I'm not so sure that's the case .

CAKE is the best option for your modest bandwidth.

I think i will just uninstall Flex then , I also see Dave archived the github for flex might he be done with the project as a whole?

Yes, FlexQoS development has ended. I’ve stopped using Asus devices after the latest struggles with the newer models and QoS. No one else has ever expressed any interest in taking over the project, so it will remain on Github for anyone to fork if they are motivated.

 

[marcd5123]

CAKE is the best option for your modest bandwidth.

Yes, FlexQoS development has ended. I’ve stopped using Asus devices after the latest struggles with the newer models and QoS. No one else has ever expressed any interest in taking over the project, so it will remain on Github for anyone to fork if they are motivated.

Hi ..Thanks for the information. I'm going to stick with FlexQos for a while yet it's to early for me to make my mind up irrespective of weather the app is still being developed or not.
I'm going to base my experience as it's happening daily and how it impacts my overhaul experience and streaming experience.
Yes on paper Cake is the one I "should" be using.. However what should be on paper correct and what I actually get in real world terms seems to be different from what some people have stated so far .

I am basing my experience as it happens daily and that is how I based my experience when I ran Cake a few weeks ago .

Having a Grade C latency in FlexQos and having a latency if A+ using Cake so obviously Cake is preforming far better on my line yet somehow I seem to for now getting a better experience with my Kodi Diggz Build with RD Streaming with Flex..So how does that figure mmmm..So for me I'm sticking with Flex for now and see how it runs for the next week and see just how "consistent" it stays !!..

Some folks have said that yes Cake is far more consistent in terms of how it performs in the the long run and Flex seems to "dwindle" after a period of time ...

So let's put that to the test in my networking environment ...if it starts becoming unstable and starts to so called misbehave then I'll switch back to Cake sooner .

Absolutely no harm trying, learning and experimenting I spent alot of money on this router the most I've ever spent so I want to try all these new things that I've never had the privilege to do so with previous routers .

So does this look correct the screenshot I have attached setup wise for flex ...

Many thanks as always to the community.

 

[Crimliar]

Previously while on VDSL2 with a 55/15 (ish) connection I found CAKE to be the best way to manage QOS when using streaming game services. Streaming from Stadia (yeah it's a little while back) at 4K it was easy to hit 25Mbps or more. Honestly, though a physical XBox never came anywhere near close enough in data use to affect anything or have any significant latency issues!
*Mileage may vary. Anyone on a connection where data rates (not connection speed) drop significantly during peak hours, is probably not going to fix things by using QOS!

 

[Kynsiä]

Ciao ..Grazie per le informazioni. Continuerò a usare FlexQos per un po', ma è troppo presto per prendere una decisione, indipendentemente dal fatto che l'app sia ancora in fase di sviluppo o meno.
Baserò la mia esperienza su ciò che accade quotidianamente e su come influisce sulla mia esperienza di revisione e di streaming.
Sì, sulla carta Cake è quello che "dovrei" usare. Tuttavia, ciò che sulla carta dovrebbe essere corretto e ciò che ottengo effettivamente in termini reali sembra essere diverso da quanto alcune persone hanno affermato finora.

Sto basando la mia esperienza su ciò che accade quotidianamente ed è così che ho basato la mia esperienza quando ho corso Cake qualche settimana fa.

Avendo una latenza di Grado C in FlexQos e una latenza di A+ usando Cake, ovviamente Cake funziona molto meglio sulla mia linea, eppure in qualche modo sembra che per ora stia ottenendo un'esperienza migliore con la mia build Kodi Diggz con RD Streaming con Flex.. Quindi, come si calcola? mmmm.. Quindi per ora resto con Flex e vedo come funziona la prossima settimana e vedo quanto rimane "coerente"!!..

Alcuni hanno affermato che sì, Cake è molto più costante in termini di prestazioni a lungo termine, mentre Flex sembra "diminuire" dopo un certo periodo di tempo...

Quindi mettiamolo alla prova nel mio ambiente di rete ... se inizia a diventare instabile e a comportarsi in modo anomalo, tornerò a Cake al più presto.

Non c'è assolutamente nulla di male nel provare, imparare e sperimentare. Ho speso un sacco di soldi per questo router, la cifra più alta che abbia mai speso, quindi voglio provare tutte queste cose nuove che non ho mai avuto il privilegio di fare con i router precedenti.

Quindi, lo screenshot che ho allegato per quanto riguarda la configurazione di Flex sembra corretto ...

Come sempre, un sentito ringraziamento alla

Hi ..Thanks for the information. I'm going to stick with FlexQos for a while yet it's to early for me to make my mind up irrespective of weather the app is still being developed or not.
I'm going to base my experience as it's happening daily and how it impacts my overhaul experience and streaming experience.
Yes on paper Cake is the one I "should" be using.. However what should be on paper correct and what I actually get in real world terms seems to be different from what some people have stated so far .

I am basing my experience as it happens daily and that is how I based my experience when I ran Cake a few weeks ago .

Having a Grade C latency in FlexQos and having a latency if A+ using Cake so obviously Cake is preforming far better on my line yet somehow I seem to for now getting a better experience with my Kodi Diggz Build with RD Streaming with Flex..So how does that figure mmmm..So for me I'm sticking with Flex for now and see how it runs for the next week and see just how "consistent" it stays !!..

Some folks have said that yes Cake is far more consistent in terms of how it performs in the the long run and Flex seems to "dwindle" after a period of time ...

So let's put that to the test in my networking environment ...if it starts becoming unstable and starts to so called misbehave then I'll switch back to Cake sooner .

Absolutely no harm trying, learning and experimenting I spent alot of money on this router the most I've ever spent so I want to try all these new things that I've never had the privilege to do so with previous routers .

So does this look correct the screenshot I have attached setup wise for flex ...

Many thanks as always to the community.

it is very strange that with Flexqos you get such a low rank in the tests have you already disabled Flow cache? if I remember correctly Flow cache is a cause of very high bufferbloat on some connections

 

[marcd5123]

it is very strange that with Flexqos you get such a low rank in the tests have you already disabled Flow cache? if I remember correctly Flow cache is a cause of very high bufferbloat on some connections

Thanks For that information..I'm not sure I have I did ask a few times and post some pictures including the above on this thread ...but as usual others so call "hijack" the thread then I had to ask a few times again about what this was originally about my tests and what I posted and needed help with FlexQos ..

Anyhow so many posts later you at least kindly acknowledge what I had kindly asked a few times above ...

Thanks I'll check that and try it again see what happens I'll report back soon.

 

[Kingp1n]

With the latest FlexQoS script.... if the FC mode is set to Auto & if you add any iptables rules (i.e. gaming or streaming rule etc...) the script will automatically turn off FC.

This will allow for traffic to flow correctly and should improve your buffebloat test score.