XR700 Bandwidth setting question

[Klesk Reaver]

Hi guys

I have been enjoying my XR700, I have a 100/40 internet speed plan (which yields 95/36 speedtests without QoS), speeds are the same with other routers I have tried so the XR700 has been working great

Something I have noticed though is that recently my speed does not change regardless of the total bandwidth setting I have set (this is the setting under the bufferbloat 'burger' icon, not the % slider)
Even if I set that to say 85 for download, I still get 95 in a speedtest

I'm wondering if this is normal behaviour? I have the "Goodput" option checked in the same section and the help for that option mentions something about speedtests

My initial concern was that QoS was not working at all, but if I set the bufferbloat setting to "Always" then my speeds do get reduced as expected to the percentage I have set (70%, roughly 66 for speed)

I remember about a month ago my speeds would never reach 95 with QoS enabled and I assumed this was just QoS reserving some bandwidth (speeds would still reach 92)
However I'm now reaching 95 both with and without QoS enabled (not including the anti-bufferbloat which continues to work as expected)

Does the Goodput option ignore your bandwidth setting for speedtests? What exactly does that setting do?

I have done a factory reset on my XR700 recently and when I was going through the initial setup wizard I did not have my internet connected on the WAN port so it was not able to detect my internet connection type and do an initial speedtest, does it matter if the router itself doesn't do this test? My internet is not a complicated setup, XR700 is directly connected to a fiber box, Internet doesn't require login (no PPPoE), IP address and DNS is Dynamic from ISP

TL;DR = Internet speedtests go higher than total bandwidth setting, is this normal?
              what exactly does Goodput setting do?

Thanks and looking forward to 3.0 especially for Application QoS and the Ad-Blocker

[Netduma Fraser]

Yes this is absolutely normal, if you use Anti-Bufferbloat but have it set to Never or even Always/When High Priority Traffic detected but have the sliders at 100% then you will still get your full speeds. Goodput enabled means speedtests will more closely align to what you expect them to be. 

[Klesk Reaver]

@Netduma Fraser

so the values you enter above the goodput option serve no purpose other than to be the basis for the % calculation for the bufferbloat?

If thats true then why not just have one option and enter the value yourself? Rather than two options? is it just easier for people to adjust a slider rather than work out on a calculator what 70% of their speeds are?

[Klesk Reaver]

@Netduma Fraser

also could you tell me what the difference is between adding a games console to the traffic prioritization rules
and setting a device type as a console in the device manager?

[Netduma Fraser]

Sorry should have been clearer, the values are important - if you put incorrect speeds in, say less than your expected but you use 'Never' for QoS, it will assume those values are your top speed and limit you to that. So they are the basis for what speeds Anti-Bufferbloat works from. I'm not sure what you mean exactly regarding that, it's much easier to just put the speeds in you expect, forget it and then just adjust the Anti-BB level.

The device types are mostly for aesthetic purposes. If it's set to a console and you add it to the Geo-Filter it will add to the Geo-Filter with the console service automatically when selected, also helps in determining what devices are playing games for Traffic Prio. With the former you're just adding a rule that can be applied to any device to prioritize any gaming ports that could be used.

[Klesk Reaver]

9 hours ago, Netduma Fraser said:

if you put incorrect speeds in, say less than your expected but you use 'Never' for QoS, it will assume those values are your top speed and limit you to that.

yeah see what i meant in my original post was that regardless of what values I enter here i was still getting my max speed of 95mbit, even if I put in 70 or 80, I still get 95 for total bandwidth. Which to me seems very odd and should not be behaving that way

Thanks for clearing the rest up though, I will try a factory reset and try and get the bandwidth option to work correctly (it used to work)

[Netduma Fraser]

Thank you for clarifying, there are only a few scenarios I can think of where I would expect that behaviour:

• QoS has been completely disabled from the Anti-Bufferbloat menu
• You've added a custom rule to Traffic Prioritization which covers the ports used for a speedtest as this would supersede Anti-BB.

Let us know how you get on after a reset.

[Klesk Reaver]

@Netduma Fraser

Hi Fraser, I'm still trying to get this to function correctly, see below for settings and speedtest

I've tried factory reset and disabling/enabling QoS again, any ideas?

[Netduma Fraser]

I would recommend re-enabling Goodput. Perhaps this has changed - I'm checking using DumaOS. Also is this causing any issues at all or more of an observation/question? If you want to throttle the connection then easier to input your speeds and use the sliders.

[Klesk Reaver]

@Netduma Fraser

more of an observation because it used to cap my speed, i havent changed firmware, is this something that can be/has been changed on a Duma server or something? What are your findings with it?

If the bufferbloat setting works as normal when ‘high priority detected’ or ‘always’ then i guess it doesnt matter that your speed can be maxed out outside of those times and in fact, probably preferable that way?

My main concern is that DumaOS is supposed to be doing something even when theres no high priority traffic (like making sure devices are still getting sticking to the bandwidth allocation amounts on the QoS ‘flower’) or something of that nature

[Netduma Fraser]

Not sure what has happened there but it is working as I described in 3.0. Yeah it probably is preferable. If you enter the max bandwidth you get then you'll never get a loss than speed, other than what you throttle to. The inconsistency is definitely fixed in 3.0, thats what I tested when you mentioned it.