Hi all !
The instructions written - "Enter your Download and Upload speeds to 90% of the results you tested in Preparation"
speadtest shows a speed of 95.5 mb
if I put in the SQM settings the speed of 85, then the speedwest will show 80
and this as I understand it is already 15% of the speed
if I put 90, then the speedwest will show 85
so the question what you need to correctly set the value of 90 or 85 ?
the instruction is not complete and I even specifically watched some videos and no one once explained this moment
i have not found this information anywhere
even ai could not answer correctly 
To the surprise of nobody... 
Have you seen the how-tosetting_up_sqm_on_a_wan_link subsection of:
Does that leave open questions?
So that's the answer to my question, it's not there.
if you know the correct answer, then tell me why I need to show the links, I can find them myself.
There is no strict correct/incorrect here, as so often the "your network your rules" logic applies.
I smell entitlement in the air, and am not amused.
Because there is no single correct value here, and to make an educated pick you need to understand a bit about the topic. Something I hope you could learn from the linked text.
P.S.: I intend to end this here, dealing with impolite/entitled folks is not my idea of quality time on a Sunday afternoon (or ever for that matter).
3 Likes
What makes you think I was rude to you?
should I be grateful for the link?
i've read this manual 20 times and didn't understand it, so I decided to ask here.
The answer is that your rules don't really suit me.
my question is essentially simple, there is nothing difficult for someone who is in the topic.
Let's say I don't know this point, so I'm asking to clarify.
I'm sure millions of users don't know this.
Your response?
That was not why I posted it, my hope was that you would understand how to optimise your shaper settings and that your initial question makes assumptions that are not strictly true.
Unsure what you mean specifically, which "rule"?
Reading the link would, ideally, have made you realise that the question is ill posed. What ever initial value you set it is just that an initial value that might be good enough for your needs, but there is no formula by which to deduce this initial setting with absolute certainty. Getting the settings right for a link typically is an iterative process.
May answer would still be the same, please read the link, ask if you have specific question.
2 Likes
I thought the answer would be much easier, since I need to study this manual for a month or even half a year
in general, as I understand it, it's not worth waiting for a quick response, thanks for that.
Bcz that's how shaping works. Plus, if you add overhead in link adaptation layer then it'll reduce bandwidth further to combat bufferbloat. Idk if you're using "ingress" keyword but if you care about strict ingress shaping without packet-backspill then you should use it.
Personally I use QoSmate (cake) with default settings without adding overhead since practically it works for me and I'm happy with the setup.
P. S. @moeller0 has been selflessly helping folks & co-authored CAKE script. Being rude to him is a new low, even for someone scraping the bottom of the decency barrel LOL. Don't tryina yap when someone tries to help you out. Many people here have contributed more value to the world than your entire post history of snarky jabs. Just chillax and be grateful. Go check QoSmate package, you'll probably enjoy it. Have a great day amigo.
4 Likes
I already figured it out myself, even made better than I planned.
And I wasn't rude, which I can't say for you.
Dude you are aggressive all the time. Do not wonder if people get annoyed by your behavior.
Your results are common, due to overhead in SQM algorithms cake/fq_codel you'll always test a little lower than your set limit, so increasing from 90 to get 85 is expected. (For example on my GL-MT6000 if I set 920,000 I get about 900Mbps with SQM fq_codel A+ ratings). This should be tweaked further by setting Overhead and MPU as the doc page says.
Ultimately your bufferbloat results are the reason why you use SQM, so if you have A+ ratings on speedtest to bufferbloat and quality then your settings worked.
I wanted to stay out of this, but the issue here is that a traffic shaper is configured using a gross rate, that is including all overhead, yet on-line speedtests only report payload throughput, which is guaranteed below the gross rate... And the recommendations that started this thread are pretty unambiguous: take a few speedtests without a traffic shaper, then set the SQM gross shaper rate to 90% of the representative speedtest result. That this results in lower payload thoughput than the gross shaper rate is the expected result.
1 Like
Always wondered why that was thanks, added that explanation to the first note on the SQM doc page.
1 Like