QoS and Ping in DOTA 2 - Please help me :)

[FreshJR]

I've been doing some more reading and does QoS really prioirtise packets? I'm trying to understand if its really ensuring packets from DOTA get processed as quickly as possible or if its just making sure the game has bandwidth.

It allots bandwidth to specific categories.

When a category is using less than it’s allotted bandwidth then its guaranteed that non of its packets will have to wait in queues/buffers. This type of behavior IS prioritization.
Once traffic categories exceeds allotted speeds and the network becomes saturated, then each category exceeding its allotted bandwidth will start queuing/delaying its packets until the speed drops down to the allotted bandwidth starting from the bottom category in WebUI traveling up.

This behavior is the reason why you don't want dota in container where it will have to fight with other traffic for allocated bandwidth. The "gaming" container is basically never queued since it both has ALOT of breathing room and is high up in the list.

For example, if dota was in the "Video Streaming" container, then both DotA && Netflix would have their packets queued to throttle bandwidth as the "Video Streaming" is fairly often limited to manage bloat. This entire behavior was explained in the QOS thread ...

The other behavior is the response time, present due to technical reasons, for QOS to take effect once the network hits saturation. As shown it's about 0.5sec. An intermittent 0.5sec ping spike should be better than a continuous ping spike without QOS.

For some reason your game doesn't reflect the dslreport results.
@Weggy have you ran pingplotter yet, pinging google for a longer ammount of time both with QOS on && off, to see a bigger picture?

Irregardless of the results, unless some devices are bypassing QOS metering, you are out of options to improve performance. Maybe try traditional with custom rules?

i noticed u have it like 8 times (4 in upload 4 in download).

8x is the correct ammount of times.
The traffic also IS showing up in gaming.

 

[Weggy]

Lastly , u did gave ur gaming console or PC a static ip address that falls under /24 cidr range correct ?
Make any sense?

Yes, this is just one machine and it's a PC. I have enabled "MAC and IP Address Binding" within the ASUS router to ensure the I.P does not change. I have not assigned a static I.P from the machine itself, I could also do this but given the DHCP will also assign the same I.P. I'm not sure if that is required. Maybe it is? Does it infact need to be out of DHCP range and "MAC and IP Address Binding" won't work?

 

[Weggy]

K I thought I'd assign the I.P. address of my PC on my PC so it didn't go through DHCP, just to see if it made a difference. Plus updated my script percentages to the below. However, I still have exactly the same issue.

    user_variables() {
        #Percent of download speed guaranteed per QOS category, change below as desired     (minimum value per section 5, sum should not be greater than 100)
        NetControl_DownBandPercent=5                    #This value can be adjust as desired        **  no spaces before or after the "=" sign **
        VoIP_DownBandPercent=20                            #This value can be adjust as desired        **                no decimals              **
        Gaming_DownBandPercent=5                        #This value can be adjust as desired
        Others_DownBandPercent=10                        #This value can be adjust as desired        #Note: New destination for unidentified traffic
        WebSurfing_DownBandPercent=20                    #This value can be adjust as desired
        Video_DownBandPercent=30                        #This value can be adjust as desired
        FileTransfer_DownBandPercent=5                    #This value can be adjust as desired
        Default_DownBandPercent=5                        #This value can be adjust as desired        #Note: Original destination for unidentified traffic, repurposed for "Gaming Downloads on ports 80/443"
 
        #Percent of upload speed guaranteed per QOS category, change below as desired         (minimum value per section 5, sum should not be greater than 100)
        NetControl_UpBandPercent=5                        #This value can be adjust as desired
        VoIP_UpBandPercent=20                            #This value can be adjust as desired
        Gaming_UpBandPercent=20                            #This value can be adjust as desired
        Others_UpBandPercent=25                            #This value can be adjust as desired        #Note: New destination for unidentified traffic
        WebSurfing_UpBandPercent=10                        #This value can be adjust as desired
        Video_UpBandPercent=10                            #This value can be adjust as desired
        FileTransfer_UpBandPercent=5                    #This value can be adjust as desired
        Default_UpBandPercent=5                            #This value can be adjust as desired        #Note: Original destination for unidentified traffic, repurposed for "Gaming Downloads on ports 80/443"
    }

As a reference I thought I'd get some images of what QoS - Traffic classification Statistics look like in different scenarios. Below are some images of those scenarios, but you can clearly see that the DOTA 2 traffic is effecting the Gaming section. Also it's worth noting that the increase in Default in the "Only Netflix running" scenario is because Steam began updating other games. I tested this and it's the case that Steam updates fall into Default.

DOTA 2 running AND Netflix running:
https://imgur.com/a/jnBYtbu

Only Netflix running:
https://imgur.com/a/8eA2mWc

Nothing running:
https://imgur.com/a/X4biRdH

Only DOTA 2 running:
https://imgur.com/a/4SpXigG

 

[Weggy]

So FreshJR kindly VNC to my computer to attmept to diagnose this issue. It appears that I get large latency spikes of over 150ms when my network load spikes. Meaning if I have a constant load or saturation, I don't have latency issues. But my Amazon Fire Stick or NetFlix spike my network load, it is not constant. Casuing these large spikes in latency.

How can this be fixed. Right now, he's not sure after trying many different things. But I'm hoping he can come up with a plan

Thanks again FreshJR for looking at the issue.

 

[Weggy]

FreshJR, I've discover that 31.55.186.177 is a BT DNS server (BT is my ISP). Meaning that with pingplotter running you can see a very simple representation of how these spikes are effecting latency. What I wanted to ask based off of this was could this be an ISP issue I'm not in control of? Meaning even if my router was configured perfectly, could BT be causing the issue by managing my data?

 

[FreshJR]

Try these settings

${tc} class change dev br0 parent 1:1 classid 1:10 htb ${PARMS}prio 0 rate ${DownRate0}Kbit ceil ${DownCeil}Kbit burst ${DownBurst0} cburst 50000b
${tc} class change dev br0 parent 1:1 classid 1:11 htb ${PARMS}prio 1 rate ${DownRate1}Kbit ceil ${DownCeil}Kbit burst ${DownBurst1} cburst 50000b
${tc} class change dev br0 parent 1:1 classid 1:12 htb ${PARMS}prio 2 rate ${DownRate2}Kbit ceil ${DownCeil}Kbit burst ${DownBurst2} cburst 50000b
${tc} class change dev br0 parent 1:1 classid 1:13 htb ${PARMS}prio 3 rate ${DownRate3}Kbit ceil ${DownCeil}Kbit burst ${DownBurst3} cburst 3200b
${tc} class change dev br0 parent 1:1 classid 1:14 htb ${PARMS}prio 4 rate ${DownRate4}Kbit ceil ${DownCeil}Kbit burst ${DownBurst4} cburst 3200b
${tc} class change dev br0 parent 1:1 classid 1:15 htb ${PARMS}prio 5 rate ${DownRate5}Kbit ceil ${DownCeil}Kbit burst ${DownBurst5} cburst 1600b
${tc} class change dev br0 parent 1:1 classid 1:16 htb ${PARMS}prio 7 rate ${DownRate6}Kbit ceil ${DownCeil}Kbit burst ${DownBurst6} cburst 1600b
${tc} class change dev br0 parent 1:1 classid 1:17 htb ${PARMS}prio 6 rate ${DownRate7}Kbit ceil ${DownCeil}Kbit burst ${DownBurst7} cburst 1600b

It should be obvious as to what section inside the script the above should overwrite.
This is the last hail mary to even out the spikes.

FreshJR, I've discover that 31.55.186.177 is a BT DNS server (BT is my ISP). Meaning that with pingplotter running you can see a very simple representation of how these spikes are effecting latency. What I wanted to ask based off of this was could this be an ISP issue I'm not in control of? Meaning even if my router was configured perfectly, could BT be causing the issue by managing my data?

So just to quell your curiosity.

It takes X ammount of hops, lets say 13, for your packet to get to its desitnation.
Pingplotter pings each of these hops as well as the destination separately and then records the results.

So 1 pingplotter interval is ~13 separate pings.

The data on top shows you the ping deviation and packet loss from each hop.
The chart on the bottom shows total ping and packet loss from the destination.

The only thing to cause concern would be if a hop along the way is experiencing packet loss or high ping deviations. This would mean that particular hop is undergoing some network issues and these stack on the end result.

For example, if you run a pingtest and saw that tje destination has 50% packet loss. If you then ran a traceroute/pingplotter and noticed a particular hop has constant 50% packet loss, then its not an issue with the destination but actually a hop along the way.

The hops in your results had no issues.

(The 50-100% packetloss when pinging the router (1rst hop) itself is nothing to be alarmed about, if that was your original question. The router is simply not responding to the individual ping requests itself since the defined interval is too short and it is treating it as a DDOS attack. Interval needs to be 1sec plus to ping the router.)

 

[Kingp1n]

8x is the correct ammount of times.
The traffic also IS showing up in gaming.

You're correct... i was using an old script on my setup.

 

[Weggy]

Try these settings

${tc} class change dev br0 parent 1:1 classid 1:10 htb ${PARMS}prio 0 rate ${DownRate0}Kbit ceil ${DownCeil}Kbit burst ${DownBurst0} cburst 50000b
${tc} class change dev br0 parent 1:1 classid 1:11 htb ${PARMS}prio 1 rate ${DownRate1}Kbit ceil ${DownCeil}Kbit burst ${DownBurst1} cburst 50000b
${tc} class change dev br0 parent 1:1 classid 1:12 htb ${PARMS}prio 2 rate ${DownRate2}Kbit ceil ${DownCeil}Kbit burst ${DownBurst2} cburst 50000b
${tc} class change dev br0 parent 1:1 classid 1:13 htb ${PARMS}prio 3 rate ${DownRate3}Kbit ceil ${DownCeil}Kbit burst ${DownBurst3} cburst 3200b
${tc} class change dev br0 parent 1:1 classid 1:14 htb ${PARMS}prio 4 rate ${DownRate4}Kbit ceil ${DownCeil}Kbit burst ${DownBurst4} cburst 3200b
${tc} class change dev br0 parent 1:1 classid 1:15 htb ${PARMS}prio 5 rate ${DownRate5}Kbit ceil ${DownCeil}Kbit burst ${DownBurst5} cburst 1600b
${tc} class change dev br0 parent 1:1 classid 1:16 htb ${PARMS}prio 7 rate ${DownRate6}Kbit ceil ${DownCeil}Kbit burst ${DownBurst6} cburst 1600b
${tc} class change dev br0 parent 1:1 classid 1:17 htb ${PARMS}prio 6 rate ${DownRate7}Kbit ceil ${DownCeil}Kbit burst ${DownBurst7} cburst 1600b

It should be obvious as to what section inside the script the above should overwrite.
This is the last hail mary to even out the spikes.

I'm afraid this doesn't appear to work either

Is there a way I can 100% confirm it is either a router issue or an ISP issue?

 

[FreshJR]

Is there a way I can 100% confirm it is either a router issue or an ISP issue?

All ISP's have bufferbloat, its just the natural result of limiting bandwidth.

Some ISP's supply modems with built in QOS mitigation to minimize bloat.
For the ISP's that don't, it is up to the user to find a solution or live with it.

 

[el pescador]

FreshJR, I've discover that 31.55.186.177 is a BT DNS server (BT is my ISP). Meaning that with pingplotter running you can see a very simple representation of how these spikes are effecting latency. What I wanted to ask based off of this was could this be an ISP issue I'm not in control of? Meaning even if my router was configured perfectly, could BT be causing the issue by managing my data?

.5 seconds always does that to my pingplotter graphs.
router cant cope for some reason.

id slow it down to the default.

this happens in my isps network too.
the spikes.

but not as bad as 200ms.

my last one did have spikes of 2000-3000ms although oddly the gameplay never represented those figures.

when comparing most of the route to the ea servers with udp big differences showed up.....even looping.

current isp oly spikes to 20-40 ms normally through one or two nodes.
this does have an effect but not a lot.
the killer with orange.es is when icmp packet loss shows up on their nodes...then gameplay is bad.

ive had 3 different isps int he last 2 years and all graphs showed different things which caused dodgy gameplay.

imo if your having issues go on their forums.
harrass them into helping.

a problem might be is they could outsource aspects of investigation and this would cause communication issues when problem solving.

i just moan on my isp forum.
they claim to be esports sponsors...i have to take the mick over this.

good luck solving issues

 

[FreshJR]

.5 seconds always does that to my pingplotter graphs.
router cant cope for some reason.

0.5second interval data are actual results. You just have more precision due to smaller time scales used.

For example, counter strike 144 tick -> interval .007 sec

Hi-Res: https://s8.postimg.cc/jlu11ynbn/results.gif
Results: https://s8.postimg.cc/ppblln8dx/bufferblaot.png

QOS is working on your router and producing results, so it is NOT a hardware issue.
If you get another Asus router, the results will be the same.