Performance on latest FW 1.0.2.86

[Jayo1177]

"Hi Netduma team,

 

Just wanted to inquire for some insight. I've recently needed to revert to the prior firmware version 1.0.2.82 from 1.0.2.86. It was depressing my performance significantly. I've made the same configuration on the prior fw. I get 1gbps and on the latest firmware I was getting depressed to 500mbps.

Upon reversion, I am not getting fully what ISP router provides. 932-950mbps on ISP router via ethernet & maximally 820 via XR1000. I wonder if there's a particular advise you might be able provide? I'm happy to advise on any settings I have currently. With the understanding from the router speed analyster over 800 will present innacurately but shouldn't a speedtest be unaffected by this?
 

Is depressed performance a known issuue on 1.0.2.86? Is there more fw updates expected? I feel like whatever vulnerability patch quoted in the latest fw could be depressing performance?"

 

I commented this on another post but figured it might be better to create a new post. I've since upgraded back to .86 & performed a factory rest. I figured maybe it would be best to be on the latest FW for support

[Jayo1177]

Hi again, 

I just wanted to offer some additonal evidence for my issue:

Here's a speedtest from speed.cloudflare:

Since the factory rest & viewing other articles on the forum I've since set nat filtering to open & configured my DNS (1.1.1.1 & 1.0.0.1) so here's some screenshots from DumaOS:

I have QoS disabled for now so nothing extraneous would be limiting performance, I guess. Appreciate guidance on this matter. I've seen naterings of a beta firmware. I wonder if this would assist my issues without having to roll back.

[Jayo1177]

You'll note that here I have improved overall speed at 633 but RTT latencies don't seem ideal & it's still depressed in comparison to 1.0.2.82.

[Netduma Fraser]

The only differences between the two are security fixes so there should not be any difference to how the versions perform, they're usually quite small so I'd be very surprised if it is the cause of the issue. The DumaOS version hasn't been touched between versions either. I don't think there will be any more updates planned unfortunately. Has QoS been fully disabled where it actually prevents you changing options on the QoS page?

The DumaOS version from the beta is the one that's in the public updates now so you're all good on that front. 

What is the model of the modem/router the XR is connected to and how have you set that to ensure all traffic flows to the XR? E.g XR in its DMZ, modem/bridge mode

Are ALL devices connected to the XR or do you still have some connected to the ISP router?

[Jayo1177]

Heya Fraser, 

Appreciate the response. That's fair. Although I can only speak to my experience & I have noted tangible differences in the firmwares. I've had a mission where I suspected it was ISP though the engineer that came to my home did the line test through the XR1000 & from their Virgin Media Hub 5. Full throughput on the router & as I mentioned half that rate from XR1000 lan port (when I opened the NAT type things did improve but it's depressed in comparison to prior firmware with secured NAT. There isn't a lot of flexibility the ISP's router fimware.

I'll apologise I'm not to familar with DMZ settings or where I might get relevant addresses and figures for them. However, ISP router is in modem mode. Previously I did have my Nvidia Shield on the ISPs router & since reconnected it to the XR when I noticed it. It hadn't changed anything that I've noticed

Also I did have QoS set to never & any Congestion Control distributions reset for this troubleshooting. 

 

 

[Netduma Fraser]

So for the Virgin hub they may have made improvements but in the past I've found modem mode to be a bit finicky. I'd suggest taking it out of modem mode then it will give the XR an IP address that you can find on the V Hub in the devices section or the same IP will show on the XR System Information page as the WAN IP in the Status box. Then put that IP in the DMZ option on the V Hub - it may be in security settings.

Disable QoS in the Congestion Control menu  fully and then re-test the speed.

[Jayo1177]

Thanks for the guidance on this one. Done as specified. I got it out of modem mode & was able to add the XR to the DMZ at 192.168.0.126.

 

Results present similarly to when the ISP router was in modem mode. QoS fully disabled:


It's a mystery to me. I'd be happy with the older firmware version but it seemed a bit less stable in other ways though speeds were better, running that in the configuration you just advised. For sure I'm happy for any other suggestions should they exist. 

 

[Netduma Fraser]

You need to click the  icon top left of the Congestion Control panel and there is an option that disables QoS fully

[Jayo1177]

 

Brought it up alright. It's shy of the ISP's stil strangely but better. Would there be any other advise? Appreciate all the help!

[Netduma Fraser]

What are you seeing on speedtest.net out of interest? With QoS disabled there isn't really much else that would effect the speed. Have you changed any other settings than what you've mentioned so far since the reset prior to testing?

[Jayo1177]

 

From the factory reset I would have only selected what's pictured previously, DNS to cloudflare. Under WAN setup DoS & Port scan, IGMP, SIP ALG are disabled. I set up my wireless of course as well. Only my phone was connected there. 

[Netduma Fraser]

Just to double check, are you doing all these tests via WiFi? What are you getting via ethernet?

[Jayo1177]

All tests discussed were done via Ethernet

[Netduma Fraser]

The only other thing I can think of that you may or may not be using is Armor, if you have it enabled then disable it please. Also no devices have been connected back to the V Hub now its out of modem mode right? As that would negatively impact the speeds through the XR.

[Jayo1177]

Armour wasn't enabled. 

I just wanted to add here to see if there's any final recommendation or path forward. I've reverted again to the prior .82 firmware & configured the same as previous with QoS fully disabled & get the following results:

 

I guess it's not so much to confound more but just to ask is there any route forward for QoS being enabled? I bought this router for these features specifically but right now there doesn't seem to be much reason to use the XR in place of the ISP router, solely. I'm not sure how much of a heightened risk I may be at for the lack of security patching. Thanks again Fraser for all the recommendations/advise up until now.

 

 

[Jayo1177]

Also no, all connections are through the XR1000

[Netduma Fraser]

Okay so speedtest.net results look like they're where they should be so that's good. Now re-enable QoS, keep Congestion Control at 100% & Never. Reset bandwidth allocation - click the button anyway even if its equal. Ensure share excess is enabled for Bandwidth Allocation in its menu  (it is by default). Ensure you don't have any manual rules added to Traffic Prioritization. Then do speedtest.net again, what speeds do you get?

[Jayo1177]

Switched QoS back on, reset distributions for download && upload. Only traffic prio is defaults as pictured (snagit artifacts with HDR capture):

 

Defiitely the best result with QoS on so far from speedtest.net:

 

Waveform Bufferbloat loaded ping has doubled though:

 

So still depressed slightly. Seems somewhat contradictory. I'm not sure what results are best to rely on. 

[Jayo1177]

For completeness sake:

These were checked

[Netduma Fraser]

I think that's the best you're going to get it seems - the difference between fully disabled/on in terms of the available router resources appears to be impacting the available speed, given nothing else has changed this will be unavoidable I think. You may get different results depending on the time of day but otherwise I think that's what you're left with.

The bufferbloat will have increased as although QoS is on it's not actually active until you lower the Congestion Control percentages/set it to Always or Auto (will only trigger when Traffic Prioritization is triggered). So to have the most available speed set CC to Auto, setup any Traffic Prio rules and when they trigger CC will trigger with whatever percentage you set, otherwise you'll get 'full' speeds like what you're currently receiving with it set as it is now. 

[Jayo1177]

Alright that's fair. 

 

In the end. Ultimately, I was able to set a 50% download limit for congestion control  achieved good scores for bufferboat consistently:

 

This is a great outcome, assuming when auto-enable is engaged I have the best of both worlds, minus the ~100mbps extra I was able to achieve with Qos completely disabled.

 

It seems couunter intuitive to me that 50% would fair better than higher values i.e., 60 - 70%. That introduced 10 - 15ms to the active download latency. Though perhaps you've explained that through your last comment already & just an indication that CC is doing it's job now.

 

 

 

[Netduma Fraser]

If you've just tried in 10% increments from 100% try in 10% increments from 95% as you might find a higher value that works better but ultimately every connection is different and so there isn't a definitive guide to what percentage works best. Some people like to really heavily throttle to 10% for example. People do tend to heavily focus on the results of that test as a mark of how good their connection is but I'd suggest going with what you feel when playing, if 50% gets good test scores but it's miserable to play on then it's not worth it.

Alternatively this is what I'd suggest doing if you have the time to really experiment: Follow this guide https://support.netduma.com/frequently-asked-questions/legacyfaqs/test-your-ping/ while downloading & start with 95% for Congestion Control (set to Always), check results, decrease by 10%, check, decrease by 10% etc, until you get to a value that is pretty good & then try 5% either side of that value to see if it can be improved. Download & Upload on Congestion Control don't have to be the same value & you may have a better experience with differing values.

[Newfoundland]

You should not need to lower that bandwidth that low to have QoS work its magic. 
 

to test your speed to see what’s going on, put your isp modem/router into router mode and check speed and if that’s low your issue is with your isp.

You should only need to lower QoS buy a small amount for a buffer, I would use ping plotter rather than that website and you can watch in real time to see the results.