PDA

View Full Version : Packet Loss


rumpleforeskin
05-06-2007, 22:11
Hi all,

a long time reader of these forums felt it was time to join, especially as i have a question.

As an avid gamer i have recently been experiancing major packet loss ( over 10% ) and its really ruining my experiance.

running winMTR i have traced the hop where the loss is occuring

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| 192.168.0.1 - 0 | 100 | 100 | 0 | 8 | 31 | 15 |
| 10.xx.xx.xx- 0 | 100 | 100 | 0 | 40 | 1032 | 1032 |
| lutn-xx.xx.xx - 1 | 100 | 99 | 0 | 32 | 141 | 15 |
| 62.252.64.101 - 1 | 100 | 99 | 0 | 31 | 109 | 16 |
| pop-bb-a-so-220-0.inet.ntl.com - 11 | 99 | 90 | 0 | 30 | 78 | 16 |
| nth-bb-b-so-010-0.inet.ntl.com - 0 | 99 | 99 | 15 | 45 | 422 | 15 |
| tele-ic-1-as0-0.inet.ntl.com - 4 | 99 | 96 | 31 | 55 | 110 | 32 |
| 212.250.14.110 - 0 | 99 | 99 | 15 | 51 | 235 | 62 |
| te4-4.telehouse-east.core.enta.net - 0 | 99 | 99 | 16 | 68 | 235 | 46 |
| th-router1.titaninternet.co.uk - 0 | 99 | 99 | 15 | 42 | 93 | 31 |
| 217.77.191.4 - 0 | 99 | 99 | 15 | 47 | 110 | 31 |
| rb-2-switch.titaninternet.co.uk - 0 | 99 | 98 | 16 | 39 | 78 | 32 |
| 85.232.32.199 - 0 | 99 | 95 | 31 | 54 | 109 | 47 |
|________________________________________________| ______|______|______|______|______|______|
WinMTR - 0.8. Copyleft @2000-2002 Vasile Laurentiu Stanimir ( stanimir@cr.nivis.com )

but who do i tell about this packet loss to get it corrected? i cannot find an e-mail address for VM and phoneing VM tech support left me on hold for 20mins before i gave up.

Is there someone who needs to know about this or will VM be aware of the problems and rectify it?

i am currently on XL package.

Thanks for listening to my wee rant
Dave

altis
05-06-2007, 22:19
Dunno about your packet loss but you could have a pic of a Shar Pei for your avatar ;)

rumpleforeskin
06-06-2007, 18:55
Dunno about your packet loss but you could have a pic of a Shar Pei for your avatar ;)

done and done :D

Incomplete
06-06-2007, 21:15
OK A few things here :)

1st your ping to your own router looks pretty high - slightly dodgy wireless?

2nd you ping to the ntl network looks too high. Could be heavy usage, could be you were downloading or uploading at the time, could be dodgy wireless or router.

3rd That trace goes through the Entanet network. The direct link between Entanet and VM has been congested for a while and probably still is. A symptom of this is packetloss.

I am curious though, 4 packets dropped to the last hop according to the sent/received columns, but 0% loss according to the packetloss column?

rumpleforeskin
06-06-2007, 23:25
The ping between my NIC and router has always hovered between 4-9, not ideal but i can live with that.

The ping between my router and the NTL network used to be under 10 but that has shot up the last week, not sure if it coincided with the packetloss i have started experiance or if its a seperate problem. I dont know what i can do to rectify this part of the problem?

and as for the Entanet network congestion i dont know what i can do about that either, i have had no luck finding the people responsible for the maintainance or smooth running of the equipment so i'm not really sure who i should be alerting to the problem

Incomplete
07-06-2007, 12:55
Probably the two are related, the high ping to VM network and packet loss.

Best speak to VM.

TheDon
07-06-2007, 15:47
I wouldn't pay too much attention to the ping to the 10.* ip it's most likely just set to deprioritise icmp traffic.
If it was actually pinging high then you'd see all of the rest of the hops in your tracert pinging higher as well (tracert's are cumulative, the ping from a to c is the ping from a to b, plus the ping from b to c so if b is pinging higher than c then it's not a true ping as that's actually impossible) but as it's the highest number in the first 5 hops it points to it just doing icmp deprioritisation (where the router delays answering it's own icmp requests in favour of routing higher priority traffic i.e. everything else)

However, that and the relatively high pings throughout does point to the ubr being overloaded, if the icmp requests are getting delayed it's because the router can't keep up with the traffic flow in real time, which would give you higher pings (you really should only be pinging 20ish within the VM network) and quite possibly some of the packet loss you see.

But basically, other than the first hop (you should only be pinging 1-2ms to your own router) there's nothing that you can do about it, other than phoning up tech support and talking to VM.

rumpleforeskin
07-06-2007, 20:57
is there a feephone number for VM internet technical support?
I dont relish the idea of paying VM for alerting them to a problem with their product

spr33
07-06-2007, 21:22
is there a feephone number for VM internet technical support?
I dont relish the idea of paying VM for alerting them to a problem with their product

If you have a VM telephone line, it's either 151 or 150 (can never remember which is which) from your VM line.

Safeman
07-06-2007, 22:12
its VM deffo m8 i have had no problems playing online games when it was ntl since VM took over its been well bad choke and loss in game it seems to happen around 6pm onwards for me i rang them they just said the network is fine wot a load of crap i have reset the modem loads of times even my internet is well slow and this is a fresh install of xp lol when i ring from my bt line takes me 20 30 mins for someone to answer every time i ring from my friends phone ntl lol 5 mins on the phone lmao money grabbers:mad:

Incomplete
08-06-2007, 00:48
I wouldn't pay too much attention to the ping to the 10.* ip it's most likely just set to deprioritise icmp traffic.

The uBRs are just 7246VXRs, not likely to deprioritise, no CoPP or RACL.

They prioritise by not having the CPU power available to answer but that shouldn't be an issue considering a uBR NPE-G1 equipped is good for thousands of customers and shouldn't be carrying more than maybe 1600 now. Maxed CPU will show packet loss at the uBR itself, but there is none the loss is further down the chain.

I'd say it's more likely to be congestion on the cable side or a signal fault, definitely needs VM to fix though as far as the high pings go :(

Sorry to the not geeky for the above :)

There's also the 'Entanet' peering issue. I think there's a packet loss problem on the other side of the peering link caused by the congestion there.

So really there are two issues here. One is the apparent oversubscription on the cable side causing high latency, the other an overloaded private peering causing packet loss.

Nice ;)

TheDon
08-06-2007, 03:31
I thought the 7200's had WFQ which could be used to set prioritisation?

If it was congestion then you'd expect to see higher pings throughout the entire trace. Although if the 1k worse case on that hop was just a one off blip that'd take the average down to 30 which would make alot more sense in relation to the rest of the trace (although still be far too high).

I'd be tempted to take the router out of the equation and then rerun the trace and see what the result is, if it's pinging up to 30 to the router then it definitely seems like there's something up with it, you'd really just expect 1-2ms.

Incomplete
08-06-2007, 08:31
WFQ is not for ICMP prioritisation as far as I'm aware?

Even in this case you can't stuff enough bandwidth down the cable (max 8 x 40Mbit) to saturate either of the uplinks (2 x GigE).

Mentioned above that the ping to the router seems a bit dodgy and I'd like to see how it looks without it as well.

Maybe that was a one off blip, perhaps routing table convergence.

rumpleforeskin
26-06-2007, 19:05
hey all

i hate bringing old topics back but i thought i would update on what has happened,

i have adjusted my wifi settings and got my ping down though it has not made much difference to my gaming as i'm still getting bad packet loss at this hop

tele-ic-1-as0-0.inet.ntl.com

usually around 8% - i have called VM twice now and they have told me they are looking into yet yet nothing has improved.

The best i can manage at the moment is finding gaming servers that do not require me going through tele-ic-1-as0-0.inet.ntl.com.

Which sadly means i dont get to play with the chaps i most enjoy playing against.

Is this situaution likley to improve itself or shoud i keep harrassing VM?