Crystal Web (NEW ISP) ADSL Feedback

Status
Not open for further replies.
Mikrotik RB750 that handels everything inc dhcp, single port billion modem, 2 tenda routers and a Billion router dhcp turned off. I can download a torrent and stream 480p youtube on 4mbps. Just need the right combination. But yeah I've got q trees, mangle rules, qos and such set up on the rb750.
 
Massive international latency

4 9 ms 9 ms 9 ms 105.233.62.1
5 32 ms 32 ms 30 ms 196.6.121.38
6 21 ms 23 ms 20 ms ct103-cpt-vl-196.ctcore.net [196.41.96.227]
7 24 ms 22 ms 22 ms vct102-cpt-gi-328.borwood.net [217.78.80.114]
8 25 ms 25 ms 26 ms vct103-cpt-gi-pa1.borwood.net [217.78.80.181]
9 29 ms 29 ms 31 ms ct201-cpt-gi-508.borwood.net [217.78.80.234]
10 170 ms 164 ms 165 ms th201-lon-po-wacs.borwood.net [217.78.88.201]
11 174 ms 179 ms 181 ms th102-lon-vl-prcera1.borwood.net [217.78.80.142]
12 185 ms 173 ms 174 ms gi-0-3-0-1-63.r02.londen03.uk.bb.gin.ntt.net [213.130.48.1]
13 188 ms * * te0-0-0-7.ccr21.lon02.atlas.cogentco.com [130.117.14.249]
14 505 ms * 504 ms be2329.ccr22.lon01.atlas.cogentco.com [130.117.49.89]
15 504 ms 500 ms * be2315.ccr22.lon13.atlas.cogentco.com [154.54.73.109]
16 511 ms * 510 ms be2276.ccr42.ams03.atlas.cogentco.com [154.54.36.58]
17 * 499 ms 507 ms be2440.mag21.ams03.atlas.cogentco.com [130.117.50.6]
 
Crystal Web said:
Perfect on our end. Tested YouTube, and two paid services. No hassles. What exactly are you trying to stream?
Click to expand...

YouTube, and Netflix.

Internet just died for about minute. Seems better again
 
Crystal Web said:
Further to this, we tested the Boks game this evening using multiple paid vendors and one free vendor. All experienced issues in XBMC such as random artifacts all over the screen, loss in picture and audio sync, and occasional buffering. Tested the same services via their web interface without issue at all other than the free stream which lost sync frequently. I honestly believe that XBMC has a sync issue with flash streams.
Click to expand...

XBMC 13 has an extra setting for the advanced settings called the readbufferfactor try setting this to about 4.

http://wiki.xbmc.org/index.php?title=HOW-TO:Modify_the_video_cache
 
Something is definitely going on...

Latency almost double, 50% odd packet loss... I can't even maintain something as trivial as a SSH session.
 
savage said:
Massive international latency

4 9 ms 9 ms 9 ms 105.233.62.1
5 32 ms 32 ms 30 ms 196.6.121.38
6 21 ms 23 ms 20 ms ct103-cpt-vl-196.ctcore.net [196.41.96.227]
7 24 ms 22 ms 22 ms vct102-cpt-gi-328.borwood.net [217.78.80.114]
8 25 ms 25 ms 26 ms vct103-cpt-gi-pa1.borwood.net [217.78.80.181]
9 29 ms 29 ms 31 ms ct201-cpt-gi-508.borwood.net [217.78.80.234]
10 170 ms 164 ms 165 ms th201-lon-po-wacs.borwood.net [217.78.88.201]
11 174 ms 179 ms 181 ms th102-lon-vl-prcera1.borwood.net [217.78.80.142]
12 185 ms 173 ms 174 ms gi-0-3-0-1-63.r02.londen03.uk.bb.gin.ntt.net [213.130.48.1]
13 188 ms * * te0-0-0-7.ccr21.lon02.atlas.cogentco.com [130.117.14.249]
14 505 ms * 504 ms be2329.ccr22.lon01.atlas.cogentco.com [130.117.49.89]
15 504 ms 500 ms * be2315.ccr22.lon13.atlas.cogentco.com [154.54.73.109]
16 511 ms * 510 ms be2276.ccr42.ams03.atlas.cogentco.com [154.54.36.58]
17 * 499 ms 507 ms be2440.mag21.ams03.atlas.cogentco.com [130.117.50.6]
Click to expand...

Appears to be an issue with Cogentco in London. Our international link there appears to be performing fine. Link to London is fine, peering on to the backup peer ntt is working fine, but the latency and packet loss seems to happen after it breaks out from our network on to cogentco. So I don't think this is related to our network at all. I can't replicate the results to other international destinations and testing the international routes appears to be fine, except for sites that require breaking out on to cogentco's network.
 
Sinbad said:
Ntt up to **** again
Click to expand...

Remove the SSL or change the SSL port. For now the shaper is going to kick in on SSL encrypted traffic after a certain threshold to attempt to preserve the international link for the time being.
 
Crystal Web said:
Remove the SSL or change the SSL port. For now the shaper is going to kick in on SSL encrypted traffic after a certain threshold to attempt to preserve the international link for the time being.
Click to expand...
Nah I'm not suffering, just commenting on the trace :-)
 
savage said:
Something is definitely going on...

Latency almost double, 50% odd packet loss... I can't even maintain something as trivial as a SSH session.
Click to expand...

I see it on the trace there, but it's isolated to cogentco traffic. I'm currently monitoring it on my end and all of the issues are on cogentco's network. It's slowly coming right though by the looks of things - stabilises to 170 for a few seconds then spikes back to 500+

You'll see that other international traces don't show the same issues, so there's nothing that anyone can really do other than wait for cogentco to fix whatever is causing the issues on their end. If it continues, then we'll report it as it may be a problem with the ntt peering link with cogentco but it looks unlikely.
 
Definitely not right.

Traffic EU -> Norris is also showing packet loss. I can't even load a web page, even though as you said, pings and packet loss seems to be better.

Ag what - time for some coffee I suppose.
 
Tracing route to www.google.com [74.125.225.115]
over a maximum of 30 hops:

1 1 ms 1 ms 1 ms myrouter.projectbismark.net [192.168.142.1]
2 1 ms 1 ms 1 ms 10.0.0.2
3 7 ms 6 ms 6 ms 105.233.54.1
4 30 ms 31 ms 31 ms 196.6.121.38
5 31 ms 31 ms 34 ms ct2-vlan610.ctcore.net [196.41.96.132]
6 56 ms 53 ms 56 ms 196.6.121.237
7 52 ms 67 ms 50 ms pk101-jhb.jhbcore.net [196.41.96.157]
8 52 ms * 53 ms 72.14.220.38
9 193 ms 198 ms 207 ms 66.249.95.8
10 226 ms 229 ms 232 ms 209.85.253.196
11 301 ms 303 ms * 64.233.175.38
12 324 ms 326 ms 327 ms 72.14.232.73
13 290 ms 296 ms * 209.85.254.239
14 286 ms * 292 ms 209.85.240.150
15 292 ms 292 ms * ord08s08-in-f19.1e100.net [74.125.225.115]
16 * * * Request timed out.
17 288 ms 280 ms * ord08s08-in-f19.1e100.net [74.125.225.115]
18 289 ms * * ord08s08-in-f19.1e100.net [74.125.225.115]
19 * 287 ms * ord08s08-in-f19.1e100.net [74.125.225.115]
20 * * 374 ms ord08s08-in-f19.1e100.net [74.125.225.115]

Trace complete.

Web pages timing out now as well
 
Cogentco traffic seems to have come right. The google traffic is odd but we can't replicate it using our DNS, although Google does route traffic its own way
 
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X