Connection test - Ping under Load advice

[Piplowe]

Hi I was just wondering if my ping under load was high in comparison to what others get. 
A jump of 4ms to 40ms does seem a large jump especially when download under load is 14ms 

Can anyone offer any insight please or ways to try and reduce the ping under load or is this only something the ISP can resolve. 

 

I also see similar results in the fast speed test when connected to the sky box 

[Exe_uz]

Double check that under QOS / Congestion control that the ALWAYS radio button is clicked

[Netduma Liam]

Good suggestion above! Otherwise could you post your current QoS configuration?

[Piplowe]

Rather bizarrely with QOS always on the under load ping for download just climbs for the duration of the test.  I tried this 3 times then set the congestion control to never, rebooted the router and still the download graph climbs. 

Also the QOS setting doesn't seem to affect the download and upload speed graph, it still show 100% not 70% which the sliders are set to.  Perhaps by design. 

[Piplowe]

Qos settings
Congestion control - auto enabled
Goodput selected

140Mb Down - 20Mb Up - Sliders set to 70% 

Traffic Prio = Duma classified games - All devices

[Netduma Liam]

The Connection Benchmark speed test supersedes the QoS configuration, just to ensure it's an accurate speedtest, so you won't see your QoS settings affect this.

Could you experiment with moving the sliders down 5% at a time and re-test in Connection Benchmark each time, then you can see which configuration provides the best results?

Are ALL devices on the network connection to the XR300?

[Piplowe]

Thanks Liam I thought that might be the case with the speed test.
Yes all devices are on the XR500.  
 

I was just wondering what sort of results other people get.  Are the upload and download under load ping times similiar? 

[Netduma Liam]

Great - thanks.

Ideally we want to be seeing the Ping Test Under Load results looking fairly similar to the regular ping test results, this will then tell us that priority traffic then won't be affected when there's congestion on the network.

[Piplowe]

Would you be able to share a screenshot so that I have something to aim for. I know the values will be different. It's just for an indication. 

 

[Netduma Liam]

19 minutes ago, Piplowe said:

Would you be able to share a screenshot so that I have something to aim for. I know the values will be different. It's just for an indication. 

 

Sure, here's mine.

 

Not perfect, but you can see the Ping test results and underload results aren't too different, that's what we're looking for.

[Piplowe]

Brilliant thanks Liam. Out of curiosity what do you get with the fast.com speed test on your main ISP router for loaded ping. 

After you run your first test you can click setting to measure loaded latency during upload 

 

Is your loaded ping much worse without the netduma ?

[Netduma Liam]

I can't actually test that currently just because of a strange network setup I've got going on, but you would expect to see a difference between loaded and unloaded without the XR500 connected, and not as much of a difference with the XR500 connected and QoS enabled.

[Piplowe]

Thanks Liam, no problem. 

I'm now seeing some strange results with the connection test where the ping just creeps up as the test goes on.  

It was fine until I changed my congestion to always on, I've tried auto and never, reset the router through the software and hard power off. 

I've been unable to recreate my first screenshot where the download was around 20 and upload was 40, any thoughts to why ?

 

[Netduma Liam]

OK and what Congestion Control configuration was this with, 70/70 still? Did you see any changes to the results after tuning this?

[Piplowe]

Hi @Netduma Liam

The rising download on the graph was happening at 70/70 but it made no difference when I turned of congestion control it was still spiking. 

The spiking has stop now and the graph is behaving like it was. However when I turn on congestion control to Always and adjust the sliders, I see no difference in the ping results. 

I've tried balanced 70/70, 50/50, 30/30
Then tried mixing it 70/30, 30/70 etc.  

• Ping is 5 ms 
• Download latency under load is 10ms
• Upload latency is under load 40ms 

Seems very odd that it's 4 times as much on upload. I'm assuming the Sky Router or ISP may be the issue. 

[Netduma Fraser]

As you're on the XR the Connection Benchmark results still need to be improved so it's unlikely you're receiving those kind of results outside the test with the Congestion Control percentages you've tried - have you had any issues that lead you to believe that are like that or are you just concerned with the results and trying to lower them? If the latter I wouldn't worry and suggest going with a Congestion Control percentage that feels good to you.

[Piplowe]

Hi Fraser, 

My gaming experience is really poor on PS5 and warzone/cold war. I've tried others to prove a point and Fortnight & Destiny have the same poor experience. 
I wonder if the reason I get a poor gaming experience is the gap between upload loaded and download loaded latency (10ms vs 40ms) .

It would be interesting to understand, how the matchmaking and lag compensation algorithms figure it all out.    
When I enter matchmaking I wonder which ping rate they use, 4ms, 10ms or 40ms and what is lag comp is being applied if join based on 4ms and my loaded latency is 40ms 

Generally most games I play in are around the 30-40ms range and even when I force them to the dedicated London servers at 4ms I still have a ping of 40ms in game. (Game processing obviously adds on a bit) 

The only thing I've found to give a bit of consistency to game play is to VPN to Denmark via HMA and geo filter to play on the UK dedicated server.  My guess is that I'm playing closer to the 40ms ping without any lag comp from 4ms or 10ms.  

I've done some testing on my mac with wired connection, I'm using ping plotter to track and I see that same pattern, 

• 4ms to bbc.co.uk 
• 10ms if I start to download something 
• 40ms if I start to upload something 

I can replicate this on both the XR500 and my Sky router.   

Ultimately as the Sky router does not offer any QOS the buffer bloat issue probably lies with them or Openreach.
I wonder how may less tickets you guys would get if the OEM ISP routers had QOS ;)  

Unfortunately Sky have tested the line, there are no faults and I can't take it any further.   
Ultimately they do not guarantee upload performance in your contract. 

[Netduma Fraser]

Really it shouldn't make a difference, if those values were correct then you would only get them when the network is severely under load - if the only thing happening was you just playing a game it wouldn't trigger those values. Is the Sky router in router mode with DMZ, bridge or modem mode at all? 

[Piplowe]

Hi Fraser,  

I see what you mean if nothing else is on the network and I'm only sending 70kbps upload and receiving around 300kbps download whilst playing the game then I shouldn't get anywhere near the loaded latency.  

I've got the XR500 in the SKY DMZ, Looking on the SKY interface and support forums I don't believe I can set the Sky Router SR204  to AP or bridge. 

In theory, If the PS5 was the only device on the sky router and in the DMZ,  When I play warzone with a ping of 4ms to the game servers, it should play without any issues or is that wishful thinking. 

[Netduma Fraser]

On the Sky router, disable the firewall settings completely, they could in theory be blocking packets. Yes in theory it would be better, however there are still a lot of variables at play, try it and see what its like, would help to narrow it down

[Piplowe]

Hi Fraser, I disabled the firewall and still get the same results.   


It seems like I'm at least half a second behind what the game is doing

Example 1 - I was shooting at someone as they ran across a doorway, I kept shooting at the doorway and hit markers appeared half a second before the person came around the corner and they killed me with the magic bullet .

Example 2 - I'll start shooting at someone, get some hit markers and then it just seems like my bullets aren't registering, after looking at the kill cam (pinch of salt) it looks like I'm dead around the time my hit markers stop registering and the game is playing catch up. 

Any thoughts? 

 

[Netduma Fraser]

It could just be the servers themselves because it's not really lag on your part, force a server with a higher ping, do you have a better experience? Some people say that you then get the benefit of lag compensation