I'm not getting much traction on my theory that there is something wrong with the Cogent route, as it seems to be fine for others. All I know is that HE gives me full speed, so as long as they restore that routing to the US I'm happy.
-
Join the MyBroadband community
South Africa’s biggest forum. Discuss, discover, and connect with thousands of members.
-
Question of the Day: Google Maps vs Waze vs Apple Maps vs physical paper maps?
Cool Ideas Fibre ISP – Feedback Thread 2
- Thread starter rpm
- Start date
- Status
I'm not getting much traction on my theory that there is something wrong with the Cogent route, as it seems to be fine for others. All I know is that HE gives me full speed, so as long as they restore that routing to the US I'm happy.
@PBCool Yesterday when tracing to 195.82.50.47 I was getting 150ms +/- now again this morning it has switched back to being 175ms. Main difference is it now uses:
192.168.11.6 > 154.0.2.94 > 154.0.4.158 > 154.0.5.18
Yesterday with the lower ms it was:
192.168.11.6 > 154.0.4.162 > 154.0.4.122 > 154.0.5.18
That x.94 range is adding the 15ms/20ms again. Can we please stop using it?
192.168.11.6 > 154.0.2.94 > 154.0.4.158 > 154.0.5.18
Yesterday with the lower ms it was:
192.168.11.6 > 154.0.4.162 > 154.0.4.122 > 154.0.5.18
That x.94 range is adding the 15ms/20ms again. Can we please stop using it?
- Joined
- Jan 11, 2016
- Messages
- 16,253
- Reaction score
- 5,582
This was the reason for the change, but have reverted until OS can make changes on capacities of the respective IPC ports.
- Joined
- Jan 11, 2016
- Messages
- 16,253
- Reaction score
- 5,582
Have you checked with our connect team?
Hi, thanks, emailed them now.
DeatheCore
Senior Member
Hi @PBCool it seems that there are a bunch of international IPs being routed through JHB instead of directly out of CPT, are you aware of this? Example:
Code:
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| router.local - 0 | 2 | 2 | 0 | 0 | 0 | 0 |
| 155.93.238.1 - 0 | 2 | 2 | 2 | 4 | 6 | 6 |
| 155.93.240.33 - 0 | 2 | 2 | 7 | 22 | 37 | 7 |
| 154.0.1.125 - 0 | 2 | 2 | 1 | 1 | 1 | 1 |
| 154.0.1.13 - 0 | 2 | 2 | 1 | 1 | 2 | 1 |
| 154.0.4.162 - 0 | 2 | 2 | 1 | 1 | 2 | 1 |
| 154.0.5.99 - 0 | 2 | 2 | 17 | 17 | 17 | 17 |
| 154.0.1.233 - 0 | 2 | 2 | 17 | 17 | 17 | 17 |
| 154.0.1.138 - 0 | 2 | 2 | 17 | 17 | 18 | 17 |
| 197.96.214.112 - 0 | 2 | 2 | 18 | 19 | 21 | 21 |
| core1b-dock-ten-0-4-3-0.ip.isnet.net - 0 | 2 | 2 | 169 | 169 | 170 | 170 |
|te0-7-0-1.ccr21.lon02.atlas.cogentco.com - 0 | 2 | 2 | 168 | 168 | 168 | 168 |
| be2791.ccr41.lon13.atlas.cogentco.com - 0 | 2 | 2 | 172 | 172 | 172 | 172 |
| be12194.ccr41.ams03.atlas.cogentco.com - 0 | 2 | 2 | 175 | 176 | 177 | 175 |
| be3457.ccr21.ams04.atlas.cogentco.com - 0 | 2 | 2 | 180 | 180 | 180 | 180 |
| 149.11.38.54 - 0 | 2 | 2 | 174 | 174 | 175 | 174 |
| unknown.hwng.net - 0 | 2 | 2 | 175 | 175 | 176 | 175 |
| 1-1.r2.am.hwng.net - 0 | 2 | 2 | 175 | 181 | 187 | 187 |
| 81.171.92.203 - 0 | 2 | 2 | 175 | 175 | 176 | 175 |
|________________________________________________|______|______|______|______|______|______|wingnut771
Honorary Master
S.Claus
Expert Member
wingnut771
Honorary Master
- Joined
- Jan 11, 2016
- Messages
- 16,253
- Reaction score
- 5,582
Yeap we switched when our carrier had issues, check again?
DeatheCore
Senior Member
Ahh that explains it, all looks good again, thanks
DeatheCore
Senior Member
Hey man, decided to test your theory re. Cogent-EU throughput vs HE-USA and I have to say, you are not crazy
and there does indeed seem to be some discrepancy with NNTP performance.I downloaded a file from ssl-eu.astraweb.com [81.171.92.201] / 20 threads, with these results:
Switched servers and downloaded the same file from ssl-us.astraweb.com [69.16.179.60] / 20 threads
So it was basically an average of 9.4 vs 11.5MB/s. Perhaps this would be more definitive if my line was faster than 100Mbps, and while it is interesting I don't really know if that is enough "proof" that Cogent has bandwidth issues or protocol shaping in place as there are probably other factors involved
I usually limit my SAB downloads to half of my line speed anyway, so it doesn't really bother me.
- Joined
- Jan 11, 2016
- Messages
- 16,253
- Reaction score
- 5,582
I can kill our Cogent transit for a day if you like?Hey man, decided to test your theory re. Cogent-EU throughput vs HE-USA and I have to say, you are not crazyand there does indeed seem to be some discrepancy with NNTP performance.
I downloaded a file from ssl-eu.astraweb.com [81.171.92.201] / 20 threads, with these results:
View attachment 679465 View attachment 679469
Switched servers and downloaded the same file from ssl-us.astraweb.com [69.16.179.60] / 20 threads
View attachment 679471
View attachment 679473
So it was basically an average of 9.4 vs 11.5MB/s. Perhaps this would be more definitive if my line was faster than 100Mbps, and while it is interesting I don't really know if that is enough "proof" that Cogent has bandwidth issues or protocol shaping in place as there are probably other factors involved
and there does indeed seem to be some discrepancy with NNTP performance.Hey man, decided to test your theory re. Cogent-EU throughput vs HE-USA and I have to say, you are not crazy
I downloaded a file from ssl-eu.astraweb.com [81.171.92.201] / 20 threads, with these results:
View attachment 679465 View attachment 679469
Switched servers and downloaded the same file from ssl-us.astraweb.com [69.16.179.60] / 20 threads
View attachment 679471
View attachment 679473
So it was basically an average of 9.4 vs 11.5MB/s. Perhaps this would be more definitive if my line was faster than 100Mbps, and while it is interesting I don't really know if that is enough "proof" that Cogent has bandwidth issues or protocol shaping in place as there are probably other factors involved
That's exactly what I see now and when I had 200/200. Cogent is extremely spikey / uneven with regards to speeds, averaging quite a few MB/sec less. HE is full speed, all the time.
DeatheCore
Senior Member
Not to worry, I have no urgency with NNTP downloads but perhaps @Seeyou and @bHOLDher might want to test further. It seems as long as there is an HE route to US servers, speeds are rock solid.
Out of curiosity though, what would happen if you had to kill the Cogent transit? Would the EU route fail over to HE? And if Cogent does have throughput issues what would the long-term solution be?
bHOLDher
Well-Known Member
To me it is not really a big problem if the downloads run at 9MB/s certain times of the day when I limit them to 11MB/s in any case, but I can imagine that the packet loss on on the cogent routes can cause some random bad experiences for other people, it might be worth a try.
Fjorko
Expert Member
- Joined
- Jan 17, 2007
- Messages
- 2,039
- Reaction score
- 960
@PBCool - What's going on ? OS 100/50 ( PLZ )
WINMTR to cisp.co.za
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| USG - 0 | 15 | 15 | 1 | 7 | 47 | 31 |
| No response from host - 100 | 15 | 0 | 0 | 0 | 0 | 0 |
| 192.168.10.5 - 0 | 15 | 15 | 113 | 129 | 160 | 126 |
| 192.168.11.6 - 0 | 15 | 15 | 98 | 116 | 159 | 159 |
| c2m-backbone.coolideas.co.za - 7 | 15 | 14 | 127 | 140 | 156 | 132 |
| 154.0.5.65 - 7 | 15 | 14 | 116 | 140 | 176 | 136 |
| 154.0.5.97 - 20 | 15 | 12 | 128 | 139 | 168 | 131 |
| 154.0.3.226 - 15 | 14 | 12 | 120 | 140 | 173 | 120 |
| 154.0.13.95 - 8 | 14 | 13 | 137 | 143 | 162 | 140 |
|________________________________________________|______|______|______|______|______|______|
WinMTR - 0.8. Copyleft @2000-2002 Vasile Laurentiu Stanimir ( [email protected] )
#COOL-20190703-270299
EDIT: Seems to be better now . . . .
WINMTR to cisp.co.za
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| USG - 0 | 15 | 15 | 1 | 7 | 47 | 31 |
| No response from host - 100 | 15 | 0 | 0 | 0 | 0 | 0 |
| 192.168.10.5 - 0 | 15 | 15 | 113 | 129 | 160 | 126 |
| 192.168.11.6 - 0 | 15 | 15 | 98 | 116 | 159 | 159 |
| c2m-backbone.coolideas.co.za - 7 | 15 | 14 | 127 | 140 | 156 | 132 |
| 154.0.5.65 - 7 | 15 | 14 | 116 | 140 | 176 | 136 |
| 154.0.5.97 - 20 | 15 | 12 | 128 | 139 | 168 | 131 |
| 154.0.3.226 - 15 | 14 | 12 | 120 | 140 | 173 | 120 |
| 154.0.13.95 - 8 | 14 | 13 | 137 | 143 | 162 | 140 |
|________________________________________________|______|______|______|______|______|______|
WinMTR - 0.8. Copyleft @2000-2002 Vasile Laurentiu Stanimir ( [email protected] )
#COOL-20190703-270299
EDIT: Seems to be better now . . . .
Last edited:
r00igev@@r
Honorary Master
- Joined
- Dec 14, 2009
- Messages
- 15,639
- Reaction score
- 14,157
- Location
- Draadloos Bantha poo doo in 4ways
But they provide the best throughput to all those dodgy porn sites?
- Status