High ping but normal download/upload speeds

[JamesMc19]

The problem started when we 'upgraded' to the newest router that my isp offers. Since then our ping has been fluctuating a lot, making it very hard for online gaming and a problem when using programs like discord. I've tried a lot of basic fixes and my isp says there's no problem with the line itself but i can't see an internal problem.

I ran a tracert to the EUW server of league of legends and had the following results:

tracert 104.160.141.3

Tracing route to 104.160.141.3 over a maximum of 30 hops

1 5 ms 5 ms 6 ms BTHUB [xxx.xxx.x.xxx]
2 * * * Request timed out.
3 * * 11 ms 31.55.185.181
4 12 ms 13 ms 12 ms 31.55.185.180
5 68 ms 16 ms 16 ms core2-hu0-12-0-1.colindale.ukcore.bt.net [195.99.127.118]
6 244 ms 54 ms 48 ms 195.99.127.66
7 39 ms 31 ms 11 ms 62.6.201.144
8 76 ms 75 ms 89 ms 213.137.183.34
9 52 ms 63 ms 55 ms t2c3-xe-2-1-1-0.nl-ams2.eu.bt.net [166.49.208.93]
10 * * * Request timed out.
11 125 ms 131 ms 117 ms 104.160.141.3

Trace complete.

If any other information is needed just ask.

 

[LinuxDevice]

Ping and traceroute use ICMP protocol. Not all hops enable this, and so a hop which times out when surrounding hops still show up are simply hops not measured. Seeing hop 2 would be useful, but timeout does not mean fault.

The time different hops require on the route are shown. Each hop has its own ping, and if the hop prior to the one with the latency is not high ping time, then it was that specific hop which slowed down...not the local router. Your third hop, where the time went up dramatically, is probably in your local neighborhood (but is not what your router talks to first). This node needs maintenance and your provider probably needs to replace wiring and/or connectors (I would guess that age and weather have compromised this particular hop). I'd show this to your ISP and tell them hop 3 is failing.

 

[JamesMc19]

Ping and traceroute use ICMP protocol. Not all hops enable this, and so a hop which times out when surrounding hops still show up are simply hops not measured. Seeing hop 2 would be useful, but timeout does not mean fault.

The time different hops require on the route are shown. Each hop has its own ping, and if the hop prior to the one with the latency is not high ping time, then it was that specific hop which slowed down...not the local router. Your third hop, where the time went up dramatically, is probably in your local neighborhood (but is not what your router talks to first). This node needs maintenance and your provider probably needs to replace wiring and/or connectors (I would guess that age and weather have compromised this particular hop). I'd show this to your ISP and tell them hop 3 is failing.

Ah ok thank you, that makes things a lot clearer. Just to be certain though, when you say 3rd hop, do you mean the explicitly 3 (e.i. 3 * * 11 ms 31.55.185.181) or the third hop that shows all ping details for that hop (i.e.5 68 ms 16 ms 16 ms core2-hu0-12-0-1.colindale.ukcore.bt.net [195.99.127.118] ).

EDIT: I also just ran the same command and the problem didn't occur until slightly later this time, would this still mean the problem is still in the same area?

tracert 104.160.141.3

Tracing route to 104.160.141.3 over a maximum of 30 hops

1 4 ms 4 ms 4 ms BTHUB [xxx.xxx.x.xxx]
2 * * * Request timed out.
3 * * * Request timed out.
4 12 ms 12 ms 32 ms 31.55.185.180
5 12 ms 13 ms 12 ms core2-hu0-12-0-1.colindale.ukcore.bt.net [195.99.127.118]
6 78 ms 90 ms 96 ms 195.99.127.66
7 94 ms 68 ms 12 ms 62.6.201.144
8 137 ms 164 ms 182 ms 213.137.183.34
9 22 ms 22 ms 25 ms t2c3-xe-2-1-1-0.nl-ams2.eu.bt.net [166.49.208.93]
10 * * * Request timed out.
11 39 ms 39 ms 46 ms 104.160.141.3

Trace complete.

 

[CaptainCretin]

Does this new router have access to error logs?

Some router chipsets dont handle bad lines gracefully, and a number of Intel ones (Puma for instance), are downright lousy on anything but a perfect line.

TBH even the 5ms 1st hop to your router looks far too long to me, my first jump on a trace (to my router) is less than 1mS, by 5mS my trace is half way to London (100 miles away as the picket flies).

 

[JamesMc19]

Does this new router have access to error logs?

Some router chipsets dont handle bad lines gracefully, and a number of Intel ones (Puma for instance), are downright lousy on anything but a perfect line.

TBH even the 5ms 1st hop to your router looks far too long to me, my first jump on a trace (to my router) is less than 1mS, by 5mS my trace is half way to London (100 miles away as the picket flies).

I have access to event logs, not error logs specifically though. I believe the reason the first hop is slightly longer is because im on a powerline but this hasnt effected my ping before so it's not really a hugely negative factor.

 

[CaptainCretin]

Does this new router have access to error logs?

Some router chipsets dont handle bad lines gracefully, and a number of Intel ones (Puma for instance), are downright lousy on anything but a perfect line.

TBH even the 5ms 1st hop to your router looks far too long to me, my first jump on a trace (to my router) is less than 1mS, by 5mS my trace is half way to London (100 miles away as the picket flies).

I have access to event logs, not error logs specifically though. I believe the reason the first hop is slightly longer is because im on a powerline but this hasnt effected my ping before so it's not really a hugely negative factor.

OK, I have never used them, so didnt know that might happen.

Back to the tracert, hop 6 is waaaaay too much, especially as you are in the UK and connecting to a server in London. My ping from the Midlands to the LOTRO servers in New Jersey are a 3rd of that hop.

Plusnet or TalkTalk??

A quick look shows a lot of complaints about PN dns routing being all over the place, and TT are the Donald Trump of ISPs.

What DNS server are you using? it might be set in the router control panel.

Try setting either Google (8.8.8.8), or OpenDNS (208.67.222.222), and see what happens.

 

[JamesMc19]

Does this new router have access to error logs?

Some router chipsets dont handle bad lines gracefully, and a number of Intel ones (Puma for instance), are downright lousy on anything but a perfect line.

TBH even the 5ms 1st hop to your router looks far too long to me, my first jump on a trace (to my router) is less than 1mS, by 5mS my trace is half way to London (100 miles away as the picket flies).

I have access to event logs, not error logs specifically though. I believe the reason the first hop is slightly longer is because im on a powerline but this hasnt effected my ping before so it's not really a hugely negative factor.

OK, I have never used them, so didnt know that might happen.

Back to the tracert, hop 6 is waaaaay too much, especially as you are in the UK and connecting to a server in London. My ping from the Midlands to the LOTRO servers in New Jersey are a 3rd of that hop.

Plusnet or TalkTalk??

A quick look shows a lot of complaints about PN dns routing being all over the place, and TT are the Donald Trump of ISPs.

What DNS server are you using? it might be set in the router control panel.

Try setting either Google (8.8.8.8), or OpenDNS (208.67.222.222), and see what happens.

I'm using BT and currently have it set to googles, could try OpenDNS though.

Changed to OpenDNS, no change:

tracert 104.160.141.3

Tracing route to 104.160.141.3 over a maximum of 30 hops

1 4 ms 4 ms 4 ms BTHUB [xxx.xxx.x.xxx]
2 * * * Request timed out.
3 * * * Request timed out.
4 11 ms 11 ms 11 ms 31.55.185.180
5 13 ms 11 ms 12 ms core2-hu0-12-0-1.colindale.ukcore.bt.net [195.99.127.118]
6 73 ms 78 ms 77 ms 195.99.127.66
7 81 ms 77 ms 76 ms 62.6.201.144
8 30 ms 46 ms 112 ms 213.137.183.34
9 19 ms 18 ms 19 ms t2c3-xe-2-1-1-0.nl-ams2.eu.bt.net [166.49.208.93]
10 * * * Request timed out.
11 80 ms 71 ms 77 ms 104.160.141.3

Trace complete.

 

[LinuxDevice]

The hop starting with sequence 3 to 31.55.185.181 completely lost some packets, and the remainder was far too long. Whatever is at 31.55.185.181 is a severe problem (or more intuitive...the line getting the packets to that node sucks).

Routers are notorious for having poor quality under load, but that node is so close to your system it isn't reasonable even for load to be the issue.

When you say "powerline", are you saying you are using networking over house wiring? This is probably something everyone should get rid of. People selling this are not doing you any favors if this is what you are using. WiFi in general can be quite terrible due to noise...power lines themselves were not designed for RF.

 

[JamesMc19]

The hop starting with sequence 3 to 31.55.185.181 completely lost some packets, and the remainder was far too long. Whatever is at 31.55.185.181 is a severe problem (or more intuitive...the line getting the packets to that node sucks).

Routers are notorious for having poor quality under load, but that node is so close to your system it isn't reasonable even for load to be the issue.

When you say "powerline", are you saying you are using networking over house wiring? This is probably something everyone should get rid of. People selling this are not doing you any favors if this is what you are using. WiFi in general can be quite terrible due to noise...power lines themselves were not designed for RF.

Yeah, networking through wiring. This is the best option for me since i dont live in the main house and wifi nor a long ethernet cable would suffice. The problem still exists in the house so even though powerline might not be the best, its still serving me well... for now.

So from here i should just contact my isp and look to get whatever the problem at that node is fixed?

 

[LinuxDevice]

You might be running into a limitation of that technology being sensitive to noise from other sources. Power lines are not shielded (unless they are in correctly grounded protective piping without corrosion between sections), and they definitely are not twisted pair. There are probably many places of failure along the route, and more than likely more than one location needs to be fixed. Typically connectors go bad or corrode, moisture seeps in, cracks too fine to see open up, so on. When it changes around like this it says that a significant section of the wiring is bad, but the source of noise is moving around. Your ISP should look into it.

A problem you will run into is that if they test it when the noise isn't showing up they won't see the issue. Someone with a real network analyzer would be required to do any serious debugging, and even then points of failure will not necessarily show up until the noise source is present. Few ISPs give their employees real network analyzers since they are thousands of dollars. Still, your ISP should be asked to check why there is so much loss. If you collect a series of suspicious traceroutes, along with the time they were taken, then this might be useful.

FYI, some ISPs charge the subscriber for repairs if it turns out the issues are in their home. Networking over power is over-hyped and they may have oversold the technology to you.