Packet Loss on NL websites/game servers

Kaishaku · 08-01-2006, 17:05

Hallo, first post here so not quite sure on the etiquette.

Basically, had ntl broadband since it first rolled out, currently on 2mb with a surfboard 3100 which I've had since the dawn of time.
I'm in Glasgow and 2 weeks ago I started getting very bad packet loss to any Dutch website or game server I connected to. German, Swiss and UK servers are always fine.

I've never had this problem before.

Phoned NTL tech support and got an Indian operator which I was prepared for. After having to explain to him what packet loss was, there followed a long conversation with many holds to try and find the problem area. After about 45 mins, I came ot the conclusion I wasn't going to get anything sorted so just accepted what he said and hung up. I know it might not be an NTL fault, but Dutch people in my clan route through the same NL ips and have no problem.

Is anyone else having this problem, or have an idea on how I should go about getting it fixed?

Here's some pathpings to a Dutch website/Unreal Tournament Server, I get PL to varying degrees on EVERY Nl website/server I connect to:

http://i44.photobucket.com/albums/f3...eenShot002.jpg
http://i44.photobucket.com/albums/f3...thping_rnl.jpg
http://i44.photobucket.com/albums/f3...g_UTserver.jpg

Thanks.

Chris W · 08-01-2006, 17:40

16-24% loss on the first hop is not good....

have you checked the signal levels to your modem? details below...

http://homepage.ntlworld.com/robin.d...ps/signal.html

Ignition · 08-01-2006, 17:48

Quote:

Originally Posted by Chris W

16-24% loss on the first hop is not good....

have you checked the signal levels to your modem? details below...

http://homepage.ntlworld.com/robin.d...ps/signal.html

Ugh and you worked tech support

Issue seems to be across AMSIX (Amsterdam Internet Exchange) between ntl and cici.nl.

This should be fixed by AMSIX, cici or ntl depending on whose issue it is, could be an issue with either cici or ntl's routers, the fibre connecting them to the AMSIX LAN, or any of the switches, copper or fibre in between!

Loss at uBR is fine as it's not repeated across ntl network.

Kaishaku · 08-01-2006, 17:53

Edited since you edited yours

Who should I get in contact with about this, or is it something that I can't have any bearing on?

Thanks for the help.

Chris W · 08-01-2006, 21:26

Quote:

Originally Posted by Ignition

Ugh and you worked tech support

Issue seems to be across AMSIX (Amsterdam Internet Exchange) between ntl and cici.nl.

This should be fixed by AMSIX, cici or ntl depending on whose issue it is, could be an issue with either cici or ntl's routers, the fibre connecting them to the AMSIX LAN, or any of the switches, copper or fibre in between!

Loss at uBR is fine as it's not repeated across ntl network.

I bow to your superior knowledge

i was only a 2nd line munkeh anyway

Chrysalis · 09-01-2006, 11:37

there are problems with some amsterdam routes currently, as for resolution I dont know I am afraid the best you can do is try and get them to log it so its escalated.

pallys · 14-01-2006, 17:33

Code:

Tracing route to quake3-06.xs4all.nl [194.109.69.95]
over a maximum of 30 hops:

  1     2 ms    <1 ms    <1 ms  pally2.mshome.net [192.168.0.1]
  2     9 ms     9 ms     8 ms  10.146.215.254
  3    10 ms     8 ms     6 ms  leic-t2cam1-a-ge910.inet.ntl.com [82.3.34.53]
  4     7 ms     7 ms     8 ms  leic-t2core-a-ge-210-0.inet.ntl.com [82.3.33.6]

  5    10 ms    18 ms     9 ms  lee-bb-a-so-220-0.inet.ntl.com [62.253.188.110]

  6    15 ms    15 ms    19 ms  pop-bb-b-so-100-0.inet.ntl.com [62.253.185.238]

  7   116 ms    44 ms    24 ms  amst-ic-1-as0-0.inet.ntl.com [62.253.188.94]
  8    28 ms    28 ms     *     ams-ix.sara.xs4all.net [195.69.144.48]
  9    31 ms    27 ms    35 ms  0.so-6-0-0.xr2.d12.xs4all.net [194.109.5.5]
 10    27 ms    25 ms    28 ms  0.so-3-0-0.cr2.d12.xs4all.net [194.109.5.94]
 11    28 ms    26 ms    26 ms  quake3-06.xs4all.nl [194.109.69.95]

I've been having the same problems with NL servers since last august (was fine before that) and has not been fixed since

terrible ping problems+packet loss

Theres been a thread on NL issues before. We kinda thought it due to traffic load with NTL in NL because at quiet hours there is no problem eg. when u get up in the morning see how the issue is then but by afternoon its gone downhill again.

My fellow UK friends on other ISPs have no issues in NL so definately is an NTL problem.

IanUK · 15-01-2006, 08:41

There looks like a peering problem with some NL routes at the moment on NTL, as Chrysalis says other ISP's don't seem to be suffering as we NTL users are so it looks like NTL's fault.

Mornings are fine, from about lunchtime onwards it all slows down until the evenings are awful, a good place to see this in action is the Giganews speed connection tester page:
http://www.giganews.com/test_connect.html

Try it at different times of the day, hopefully NTL will fix this, but I'd guess they won't hurry themselves as they never seem to spot peering problems until its too late.

RichieUK · 15-01-2006, 09:14

I've been having the same issues for a while now on an online game called 'Planetside'. Through the day i get a loss of around 0.2%, which is fine, but from about 6 or 7pm this goes upto anything from 7-15% loss

It's just NTL aswel because people in my squad on BT etc are getting 0.0% loss all the time

IanUK · 15-01-2006, 17:26

Speeds are awful to Giganews currently and also to that quake server you mentioned, surely someone at NTL should be monitoring these terrible peering points, its been like this for weeks now.

I had the same problem with Cogent and the only way it got fixed was by going directly to Cogent and moaning, telling NTL about it did nothing and it looks like the same is happening here.

Not much point advertising these 10mb speeds when your peering is so stuffed that even 1 meg is a waste of time.

Someone at NTL isn't doing their job properly it seems...again

Chrysalis · 15-01-2006, 17:34

cogent isnt really fixed its just been worked around, the routing is going through germany and france and then to usa which is far from optimal and I get at least 1 drop a day, I wake up and a ssh session (going over cogent) will be disconnected. When over london the drops arent there but cogent in london is saturated.

IanUK · 15-01-2006, 21:31

Cogent has been fine here since the fix, good speeds and reliability.
I'd guess this is location dependent again, some routes seem to go through London, some Europe, both are fine here though currently.

NL routed stuff is rubbish though, NTL don't seem to have any kind of monitoring of peering at all, I've never known an ISP be so rubbish at something that should be one of their core strengths, both BT and AOL are fine currently everywhere I try, even at peak times.

NTL just seem to keep boosting speeds but not bothering about the basics to keep it all working...

pallys · 16-01-2006, 10:33

so how can we get this escalated and fixed ?

I know this problem has existed since around august 2005.

is it a problem with ntl's peering on amsterdams exchange amsix ?

i found these details on NTL on amsix

https://www.ams-ix.net/connected/

https://www.ams-ix.net/connected/sho...68549aa236ce7b

IanUK · 16-01-2006, 11:42

You could try emailing those contacts you found directly, ringing support does nothing for these sorts of issues, wouldn't hurt to try, also raise a responsetek from this site.

Chrysalis · 16-01-2006, 13:19

well the cogent drops are out of ntl's control of course that is cogent themselves, here is an example of some poor cogent routing however, the destination is .uk.

1 25 ms 74 ms 47 ms UBR [10.8.143.254]
2 36 ms 20 ms 21 ms leic-t2cam1-a-ge914.inet.ntl.com [82.3.35.149]
3 41 ms 83 ms 22 ms leic-t2core-a-ge-220-0.inet.ntl.com [82.3.33.10]
4 25 ms 10 ms 90 ms lee-bb-a-so-220-0.inet.ntl.com [62.253.188.110]
5 40 ms 91 ms 44 ms nth-bb-b-so-000-0.inet.ntl.com [62.253.185.101]
6 17 ms 22 ms 14 ms nth-bb-a-ae0-0.inet.ntl.com [62.253.185.117]
7 45 ms 32 ms 97 ms fran-ic-1-as0-0.inet.ntl.com [213.105.174.86]
8 68 ms 30 ms 43 ms po1-0.core01.fra03.atlas.cogentco.com [80.81.192.63]
9 45 ms 42 ms 38 ms p3-0.core01.ams03.atlas.cogentco.com [130.117.0.145]
10 70 ms 32 ms 32 ms p5-0.core01.lon02.atlas.cogentco.com [130.117.1.225]
11 166 ms 68 ms 50 ms oh-tele.demarc.cogentco.com [130.117.241.194]
12 35 ms 45 ms 73 ms web2.hostingfreak.net [195.242.215.140]

again I believe this may not be ntl's fault since cogent london does appear to be congested I know nildram are also avoiding cogent london but they only route through france and not germany.

(yes my pings do suck but thats another topic)