CakeQOS CakeQOS-Merlin

[archiel]

I am planning to install in the morning (when the network is quiet and complaints from the family can be avoided) and have a question around overheads

I am connected to Sky Broadband in the UK (80/20) VDSL2 using a Vigor 130 modem (1483 Bridged IP LLC) connected to the AX88U and using DHCP option 61 to connect. Should I treat this as pppoe-ptm or does having the modem first mean I should be using a different overhead setting.

 

[Kingp1n]

this is what i use for my comcast:

[1]  --> Download Speed             | [285 Mbit]
[2]  --> Upload Speed               | [0 Mbit]
[3]  --> Queue Priority             | [besteffort]
[4]  --> Extra Download Options     | [docsis dual-dsthost]
[5]  --> Extra Upload Options       | [docsis ack-filter dual-srchost]

note that i use zero for upload because my local node upload changes during the day (from 2Mbps to 20Mbps).
it you have stable max speed on your connection, a non zero limit gives better results than using zero
i use ack-filter on the upload to help reduce bloat there. not needed on download since i have enough bandwidth there; using besteffort because available categorization is poor for diffserv4, apparently (i'm conforming to info found on bufferbloat.net and on this thread). also using 285 on the download because that appears to be a cake hardware limit on ax88.
but note, some people use zero for both with good results. depends on your connection.

start with 95/9 and check results
(this is, use 95% of your speeds, as measured by spdMerlin with cake/qos off)

Thanks for the info.

I am using the below settings under extra:

[1]  --> Download Speed          | [100 Mbit]
[2]  --> Upload Speed               | [4 Mbit]
[3]  --> Queue Priority             | [besteffort]
[4]  --> Extra Download Options | [docsis ack-filter dual-dsthost]
[5]  --> Extra Upload Options       | [docsis ack-filter dual-srchost]

So far, I'm having no issues but just wondering if i should remove 'ack-filter' under download...is there any significant difference in using the following "extra option" for download for my speeds (100/5mbps).

'docsis ack-filter dual-dsthost' vs. 'docsis dual-dsthost' ?

 

[ugandy]

Thanks for the info.

I am using the below settings under extra:

[1]  --> Download Speed          | [100 Mbit]
[2]  --> Upload Speed               | [4 Mbit]
[3]  --> Queue Priority             | [besteffort]
[4]  --> Extra Download Options | [docsis ack-filter dual-dsthost]
[5]  --> Extra Upload Options       | [docsis ack-filter dual-srchost]

So far, I'm having no issues but just wondering if i should remove 'ack-filter' under download...is there any significant difference in using the following "extra option" for download for my speeds (100/5mbps).

'docsis ack-filter dual-dsthost' vs. 'docsis dual-dsthost' ?

from what i found online, links posted on the thread, unless you DL saturates often, you don't need the ack-filter there. your UL is more likely to saturate so you should have it there.
also, if your connection UL/DL changes during the day (as in going below your chosen limits), you might want to test using 0/0

 

[Kingp1n]

from what i found online, links posted on the thread, unless you DL saturates often, you don't need the ack-filter there. your UL is more likely to saturate so you should have it there.
also, if your connection UL/DL changes during the day (as in going below your chosen limits), you might want to test using 0/0

Thanks!!! I removed 'ack-filter' from the extra options under Downloads.

I can't say enough how much difference I'm seeing at home while using 'cake' with comcast and my sucky speeds haha. I have 3 firesticks running wireless in a 3 story home (2 of them running Kodi with livetv playing (Arsenal vs. LC), all playing in HD, 1 PS4 (wireless) playing CoD all happening without any lag!!! I also 15+ IoT devices connected without any issues.I've been struggling to get this setup in the last month and I can finally say, I found it!!! Awesome job to everyone involve here! You guys rock!

 

[Wade Coxon]

Is this the best way to test internet latency/jitter ping all that matter also for online gaming?

Or is connmon and spd Merlin better?

I don't use any tool on the router for that.
The DSLReports speedtest is what I use. I figure this is a valid way to test as it is actually initiating the testing from the PC that will actually do the gaming.

DSLReports will provide you with a quick and easy "rating" for Bufferbloat, but if you look at the results page you will be able to see the actual average figures for ping latency under load v latency at idle.

 

[raion969]

On speedtest.net i get around 95-97 on download and in DSL reports i get around 104-106 So what is the actual real number?

 

[Wade Coxon]

On speedtest.net i get around 95-97 on download and in DSL reports i get around 104-106 So what is the actual real number?

Those tools will report some munged version of an average value or the last obtained value, and won't really give you a good idea of your link's actual capacity. That's especially the case if you have any other traffic on your network at the time.

Leave the Asus "Traffic Monitor" page open in Real-time mode when you run those tests.
Use the maximum steady-state value reported on that graph. Mouse-over the graph at a nice high flat section and that's your value in kB/s. Multiply that by 8 to change kB/s to Mbit/s.

Knock 5% off that figure and use that as your QoS limits. (separate limits for upload and download of course).

See if DSLReports then reports any improvement in bufferbloat. Keep adjusting down in 1% increments until you find the point where your bufferbloat loaded ping values are less than 10ms above your idle pings.

Depending on which QoS solution you are using, you may need to wait a little while for the rules to kick in before you re-test. In the old FreshJR script, you had to wait 5 minutes. I understand that is vastly reduced with the FlexQoS script, but don't know what the value is. With Cake, you don't need to wait after setting the QoS limits, but unfortunately that won't be an option for your router.

 

[Wade Coxon]

A few days ago, I was having a play around with the DSCP marking script listed in this OpenWrt thread to see if it would do anything on AsusWrt.
https://forum.openwrt.org/t/using-dscp-for-qos/14252/6

I obviously replaced "$IPT" with "iptables", and then taking a punt, I changed the forwarding references of eth1 with eth0, and the ingress reference to eth1 with ifb9eth0.

Running it in stages started well, but quickly started throwing out errors with chains already existing and other oddities that I lack the knowledge to diagnose any further Any attempt I made to find the already-existing chains didn't seem to show anything.

Anyway, does someone smarter than me know if this approach would even work with our routers? I get the impressing that iptables based DSCP marking will be fairly slow, but maybe this can be explored as a proof of concept?

 

[L&LD]

On speedtest.net i get around 95-97 on download and in DSL reports i get around 104-106 So what is the actual real number?

Both.

They are using different servers and testing methods.

Use both along with fast.com and others to get a broad idea of what your speeds actually are for each respective test.

 

[Kingp1n]

The nice part is that: Runner: Enabled - Flow Cache: Enabled
Makes for a fast experience.

@skeal...i looked at my settings while using cake and both Runner/Flow Cache are disabled. Could this be due to using AIProtection? Is there a way I can enable these features? It shouldn't matter with my speeds but just curious now why mine are showing as disabled.

 

[Wade Coxon]

@skeal...i looked at my settings while using cake and both Runner/Flow Cache are disabled. Could this be due to using AIProtection? Is there a way I can enable these features? It shouldn't matter with my speeds but just curious now why mine are showing as disabled.

They're not enabled. When @skeal posted that, the acceleration values were not being properly reported in the web interface. The script has now fixed that.

 

[rgnldo]

One week with a stable connection. Great.

 

[maghuro]

One week with a stable connection. Great.

View attachment 24603

And they say internet in Brazil is slow...
What can we say here in Portugal?
But cake really does its magic

 

[rgnldo]

And they say internet in Brazil is slow...
What can we say here in Portugal?
But cake really does its magic

There are demanding infrastructure and IT professionals. Like me. Generating technology (open source) is rewarding. I am porting Cake-QOS to an ISP.

 

[maghuro]

There are demanding infrastructure and IT professionals. Like me. Generating technology (open source) is rewarding. I am porting Cake-QOS to an ISP.

Wow!
A balancer like cake being used on ISP side... It's a dream come true

 

[Kingp1n]

@maghuro are this test being done wireless or wired? It seems its from a mobile phone so it should be wireless! Pretty good!

I keep getting B/C bufferbloat (which is already a big improvement for me haha) scores with settings below:

[1] --> Download Speed | [90 Mbit]
[2] --> Upload Speed | [4 Mbit]
[3] --> Queue Priority | [besteffort]
[4] --> Extra Download Options | [docsis dual-dsthost]
[5] --> Extra Upload Options | [docsis ack-filter dual-srchost]

Quality & Overall is always showing as A/A+. Any suggestions I can try to get A bufferbloat scores? My paid speeds are 100/5 mbps but I normally get 115/6 mpbs on actual speed tests.

Are my limits set to low?

 

[maxsteel]

Connecting to ISP fiber modem via IPTV settings, PPPoE from RT-AC86U. Full speed is 300/80.
Any recommended changes either in speed values or Extra dl/ul options? is it possible to add overhead value in cake? Thanks in advance.

 

[brummygit]

@maghuro are this test being done wireless or wired? It seems its from a mobile phone so it should be wireless! Pretty good!

I keep getting B/C bufferbloat (which is already a big improvement for me haha) scores with settings below:

[1] --> Download Speed | [90 Mbit]
[2] --> Upload Speed | [4 Mbit]
[3] --> Queue Priority | [besteffort]
[4] --> Extra Download Options | [docsis dual-dsthost]
[5] --> Extra Upload Options | [docsis ack-filter dual-srchost]

Quality & Overall is always showing as A/A+. Any suggestions I can try to get A bufferbloat scores? My paid speeds are 100/5 mbps but I normally get 115/6 mpbs on actual speed tests.

Are my limits set to low?

When running the tests your objective is to have the router (your internet connection assuming you have a standard topology) as the pinch-point (slowest link) in the connection from your device all the way to the speedtest site. This means that it you test over WiFi (or even just 100Mbps ethernet) you run the risk of the wifi being the pinch-point and therefore bufferbloat can be introduced at a different point in the journey not benefitting from cake. Cake is applied to the Internet connection on the router, so while you are unlikely to have a problem with your upload, your download traffic could queue in the router waiting to reach your client device. If you normally see 115/6 in speed tests under the same conditions this is unlikely to be an issue for you however.

A couple of things to think about:

1. Is DOCSIS the correct setting for you?
2. It might be worth changing the download to 50 and upload to 2 and see what results you get - the speedtest should be limited to the values you set but I would expect excellent bufferbloat, if not you need to look at why not. If it does you can start to increase and find your sweet point. I found that sometimes a router reboot (or even a stop and restart cake) could be beneficial in the early days of testing after making lots of changes. I'm not sure why (maybe it was all in my head) but picking some settings, rebooting your router and waiting 10 minutes for everything to settle might be beneficial.

 

[ttgapers]

or Extra dl/ul options? is it possible to add overhead value in cake? Thanks in advance.

See for overhead based on connection type. Added in extra dl/ul: https://openwrt.org/docs/guide-user/network/traffic-shaping/sqm-details#sqmlink_layer_adaptation_tab

 

[Kingp1n]

When running the tests your objective is to have the router (your internet connection assuming you have a standard topology) as the pinch-point (slowest link) in the connection from your device all the way to the speedtest site. This means that it you test over WiFi (or even just 100Mbps ethernet) you run the risk of the wifi being the pinch-point and therefore bufferbloat can be introduced at a different point in the journey not benefitting from cake. Cake is applied to the Internet connection on the router, so while you are unlikely to have a problem with your upload, your download traffic could queue in the router waiting to reach your client device. If you normally see 115/6 in speed tests under the same conditions this is unlikely to be an issue for you however.

A couple of things to think about:

1. Is DOCSIS the correct setting for you?
2. It might be worth changing the download to 50 and upload to 2 and see what results you get - the speedtest should be limited to the values you set but I would expect excellent bufferbloat, if not you need to look at why not. If it does you can start to increase and find your sweet point. I found that sometimes a router reboot (or even a stop and restart cake) could be beneficial in the early days of testing after making lots of changes. I'm not sure why (maybe it was all in my head) but picking some settings, rebooting your router and waiting 10 minutes for everything to settle might be beneficial.

@brummygit thanks for the response! I'll test those speeds with my setup.

As far as "docsis" being used...im using due to having a cable ISP (comcast). Any other recommendations I can try under the extra options?