×  Introducing Nighthawk RS700 WiFi 7 (BE) Router. For more details, please visit Nighthawk Tri-Band WiFi 7 Router.
Reply

XR1000 - Bufferbloat - False negative?

TTedd
Aspirant

XR1000 - Bufferbloat - False negative?

Hello all, I have been running the XR1000 for the past week.  Enoutnered a few bugs/issues, but real reason here is to query others on their bufferloat (DSL reports) experience.  I am seeing a degradation in performance vs a comparable Asus unit AX11000.   Maybe it is false negative (as this is not gaming traffic), but what are you seeing?  Below outlines my current experience.

 

Setups tried all with similar results...

Firmware - latest .44

Cable connection 400/35

Bandwidth speed set to 285/18 - well within limits

Device doing the testing = PC defined as a device type = computer.

Connection is hardwired 1GHz ethernet 1ms ping to router / no packet loss, etc

Congestion traffic tried both always on and auto @ith congestion limits @ 70-90%..went as low as 50% no real change

QOS - devices (3) @each set @ 33% and by traffic 10@10% default

Share excess on and off for downloads and uploads.. both and one at a time e.g. on/on, on/off, off/on, off/off

Traffic prioritzation default =on, duma classified games and also adding the device running the test 

For comparison Asus router has QOS turned on adaptive, bandwidth limit set to 300/18

 

DSL multiple tests in with minimal house side changes - both show same throughput 280-285/17-18Mbps

XR1000 Bufferbloat A-C Quality A vs. Asus Bufferbloat A+ Quality A+

 

XR1000 observations

Download will spike bufferbloat to 80-100ms during ramp up and reduce down to 15-25ms with max steady state data transfers

Upload will spike to 280-350ms during ramp up and reduce down to 30-35ms with max steady state data transfers

note: Now in Duma OS built in bench test I see good ping stability idle vs upload vs download. +/- 3ms variation

 

Asus AX11000 observations

Download will spike bufferbloat to 10-24ms during ramp up and reduce down to 1-6ms with max steady state data transfers

Upload will spike to 20-35ms during ramp up and reduce down to 5-8ms with max steady state data transfers

 

Issues I have run into while doing all sorts of reconfigs/reboots...

1. Guest network 2.4G was broadcasting (on) while turned off in settings - had to do a factory reset to resolve this. 

2. Using Edge as browser ran into a crash failure after renaming multiple (15+) devices in duma os device manager - issue occurred when trying to go to network monitor - something about a script block 192.168.1.1 (sorry never screen captured exact wording).  Had to reboot modem to recover from this issue.

3. QOS traffic type... switching between devices and traffic type results in 10 traffic types each being assigned 11% bandwidth.  Used reset distribution to fix the 11% issue on both upload and download.

4. At this time I was also seeing express sharing set to off act like it was turned on. Had to go in and turn it ON and then back OFF to get it to apply the hard limits by device - which was the mode I was trying validate at the time.  Before resolving issue throughput was going to max set in bandwdith limits 285Mpbs vs what should have limited to 95Mbps (3 device @ 33% each) for downloads.  Same behavior was seen on uploads too.  Yes I am sure it was set correctly on both download and upload.

5. Duma classified games - traffic prioritization was not being applied (red button icon remained off) when an XB1X hardwired to an ethernet port was actively running games actively, device is set as an XBOX in device manager.  Added device to traffic prioritzation to force it to prioritze traffic.  Did not see this on prior XR version units.

 

Suggestion on future QOS options

1. In device mode ... provide option to have set excess share turned off by select device and rest act as pool of devices with share excess turned on.  This is really useful in a high count device situation where you want 2-3 devices to be explicitly bandwidth protected and the rest act as pool... pseudo work around for this is create an aggregation unit ahead of the modem e.g. another router - hence the pool is seen one device.  But this defeats the point of having WIFI on the main router... the things we will do/try to get best gaming experience.

 

Message 1 of 5
Netduma-Fraser
NetDuma Partner

Re: XR1000 - Bufferbloat - False negative?

I wouldn't really give much credence to DSLReports to be honest. That is why we added Connection Benchmark, I would recommend you try different CC settings to get the Ping Under Load test results down as much as possible and then it should be as optimized as possible. We have had that suggestion before so likely we will add it in the future.
Message 2 of 5
TTedd
Aspirant

Re: XR1000 - Bufferbloat - False negative?

Thanks for response.  Having not had much run time on the unit it is hard to assess actual performance in the real world, hence my post.  The results though being so destinctly and repeatably different raised doubts and questions.  At this point I will trust the internal benchmark (which is really nice added feature) and move forward 🙂

 

 

Message 3 of 5
TTedd
Aspirant

Re: XR1000 - Bufferbloat - False negative?

Connection benchmark gives me A+ / A / A+.  Being on a copper cable setup this seems decent.  I have noticed ping rdeviates little bit based on an upload or download.  When CBenchmark runs thrid test I see results like idle=25ms, upload =28ms, download=22ms... minimal jitter variations.  Download is always reporting lower than idle.

Message 4 of 5
Netduma-Fraser
NetDuma Partner

Re: XR1000 - Bufferbloat - False negative?

Glad you like the feature! Those results seem really good so give them a try via game and see how you get on!
Message 5 of 5
Top Contributors
Discussion stats
  • 4 replies
  • 1204 views
  • 2 kudos
  • 2 in conversation
Announcements

Orbi WiFi 6E