and maybe whole europe?


Tried a variety of sites, all are REALLY SLOW - almost down, if I surf them from germany's biggest provider t-online. This is by far the biggest provider in europe with about 20 mio users, so it does concern a lot of people.

This begun yesterday. Before it was a nice speed but since yesterday awfull SLOW.

I figured it out on my 2 rackshack servers first. SSH sessions are almost impossible. And I get transfer rates of below 1 KBit. This is on DSL and since yesterday I would get 20 KB and more. I figure rackshack delivers europe through cogent, canada through time warner and us through verio? On their regular boxes that is, those for $99 etc..

The rackshack.net homepage is fast though

And it's not only rackshack, ALL sites that I try, that are hosted on cogent bandwidth, are almost worthless to run for the surfers on t-online

Seems like cogent dropped a peering point yesterday?

Because with other, smaller providers I STILL get everything fast.
So it is the fault of t-online then, and not cogent?



I hope it doesn't stay that way, it's really dissapointing. It's not like this a small market

I've got no problems with cogent at all. And I'm in Holland ...:eek:

Yes it has to do with the combination of t-online & cogent I guess. That combination is so slow

I can still surf fast with other small dial in providers.

But t-online is the cheapest, biggest and (used to be) best provider here, that is why I am concerned. I am thinking about the german surfers that will hit the stop & back buttons since yesterday when surfing my sites


Such things apparently (slow connection t-online to us) happened once in a while but just for minutes or maybe an hour. This one lasts for almost a whole day now

Just curious if this is slow for you (has test files):

http://bassboy.net

(It's on a cogent line)

I tried the 11 MB file for ~ 2 minutes. Started at 178 Byte/s and dropped to 78 Byte/s later.

Webhostingtalk for instance, is coming up fast as ever


Well, I'm keeping fingers crossed that someone somewhere is working on it right now, hopefully...

Contacting t-online wouldn't make sense as it almost impossible to get through to them even with billing issues, unless you threaten them with legal action ;-)

One of our clients is in Finland and has problems with cogent as well. So, maybe it's not your ISP.

Have you done a traceroute already ?

Routenverfolgung zu 64.246.28.78 über maximal 30 Abschnitte

1 55 ms 55 ms 55 ms 217.my ip
2 55 ms 55 ms 55 ms 217.237.157.94
3 151 ms * 165 ms Vienna-gw12.USA.net.DTAG.DE [62.156.131.190]
4 151 ms 152 ms 151 ms p4-1-1-0.r00.stngva01.us.bb.verio.net [129.250.9
.197]
5 151 ms 164 ms 151 ms p16-1-0-0.r01.mclnva02.us.bb.verio.net [129.250.
3.157]
6 165 ms 179 ms 178 ms p16-0-0-0.r00.atlnga03.us.bb.verio.net [129.250.
2.49]
7 179 ms 164 ms 179 ms ge-1-2-0.r01.atlnga03.us.bb.verio.net [129.250.5
.246]
8 193 ms 192 ms 192 ms p4-1-3-0.r00.hstntx01.us.bb.verio.net [129.250.2
.45]
9 178 ms 206 ms 192 ms ge-1-1-0.a03.hstntx01.us.ra.verio.net [129.250.3
0.216]
10 179 ms 192 ms 192 ms ge-0-0-0.a03.hstntx01.us.ce.verio.net [128.241.2
.102]
11 * 371 ms * tayhou-223-36.ev1.net [207.218.223.36]
12 * 357 ms * 64.246.28.78
13 371 ms



Routenverfolgung zu 64.246.28.78 über maximal 30 Abschnitte

1 55 ms 55 ms 55 ms 217.my ip
2 55 ms 55 ms 55 ms 217.237.157.94
3 151 ms 151 ms 152 ms Vienna-gw12.USA.net.DTAG.DE [62.156.131.190]
4 151 ms 151 ms 151 ms p4-1-1-0.r00.stngva01.us.bb.verio.net [129.250.9
.197]
5 151 ms 151 ms 151 ms p16-1-0-0.r01.mclnva02.us.bb.verio.net [129.250.
3.157]
6 164 ms 179 ms 178 ms p16-0-0-0.r00.atlnga03.us.bb.verio.net [129.250.
2.49]
7 165 ms * 179 ms ge-1-2-0.r01.atlnga03.us.bb.verio.net [129.250.5
.246]
8 192 ms 192 ms 206 ms p4-1-3-0.r00.hstntx01.us.bb.verio.net [129.250.2
.45]
9 192 ms 193 ms 192 ms ge-1-1-0.a03.hstntx01.us.ra.verio.net [129.250.3
0.216]
10 192 ms 192 ms 192 ms ge-0-0-0.a03.hstntx01.us.ce.verio.net [128.241.2
.102]
11 * * * Zeitüberschreitung der Anforderung.
12 371 ms 371 ms 370 ms



that last line says "time limit exceeded"

so it ain't cogent? Strange as it seemed that I get it with sites hosted on cogent BW (when I made tracert from a visualroute server in germany instead of my own IP)

sniff sniff........... I did all this work for nothing?


http://webhostingtalk.com/showthread.php?s=&threadid=48075

Originally posted by ClusterMania
sniff sniff........... I did all this work for nothing?


http://webhostingtalk.com/showthread.php?s=&threadid=48075

:D

But maybe my problem isn't cogent related after all? Anyway, I'm confused, let's proceed at that other thread hehe

:cartman:

most cogent lines are fast from austria. on chello cable.

i tried that test file, it burst to 388k, then settled in at around 70.

not bad at all really.

Cogent started the integratin with PSINet earlier this week. They acknowledge some problems and are working through them this weekend. Hopefully, by Monday or Tuesday it will be MUCH better , as even better than before.

Hi Erich,

I see you are stil getting problems, was it you I was talking to on rackshacks support page yesterday?

Hope you get it sorted.

Brian

i was traceing out of rackshack having a look at what cogent were doing and seeing if it was affecting our servers there and being in the chat help room at the same time there were quite a few comments about people having network problems, turned out not to be cogent at all, verio seemed to be having a few problems a hop or two out of rackshack and rackshack 'seemed' to be having a minor router problem. not being an expert i can only report what i saw and heard but it did all seem to sort itself out.
cogent has got a lot better to europe. they also seem to have a quirk, they dont seem to support fast connections but can handle many more slower ones i.e. i can download a file at 100 kbps over cogent on a link that should go a lot faster, but i can download 5 files at the same time all at 100 kbps. ive tried this from several sources so its cogent alright. they arent the best to europe but they seem to be getting better (maybe even upgraded their 33.6k transatlantic link to 56k?). maybe t online isnt great at peering?

I'm having difficulties as well accessing my RS server from Canada. Actually, it's gotten progressively worse throughout this morning.

whats your trace? is verio involved?

hey Erich,

here's another guy from Germany, online with T-Offline ;-)
well i think i found the reason ....
just have a look at this:

[root@gw root]# traceroute bassboy.net
traceroute to bassboy.net (66.28.147.105), 30 hops max, 38 byte packets
1 62.225.255.225 (62.225.255.225) 28.580 ms 28.312 ms 28.783 ms
2 62.225.255.226 (62.225.255.226) 29.180 ms 29.476 ms *
3 Vienna-gw12.USA.net.DTAG.DE (62.156.131.190) 118.368 ms 119.025 ms 128.461 ms
4 sl-gw20-rly-2-0.sprintlink.net (160.81.255.9) 180.636 ms 135.254 ms *
5 sl-bb20-rly-3-2.sprintlink.net (144.232.14.29) 153.320 ms 121.473 ms 123.025 ms
6 sl-bb22-sj-5-1.sprintlink.net (144.232.9.125) 173.053 ms 222.110 ms 181.431 ms
7 * sl-bb20-sj-15-0.sprintlink.net (144.232.3.166) 172.766 ms 173.513 ms
8 sl-gw8-sj-9-0.sprintlink.net (144.232.3.110) 174.544 ms 175.414 ms 174.070 ms
9 sl-cogentcomm-7-0.sprintlink.net (144.228.110.122) 354.118 ms * 440.806 ms
10 p6-0.core01.sfo01.atlas.cogentco.com (66.28.4.145) 354.142 ms * *
11 p14-0.core01.ord01.atlas.cogentco.com (66.28.4.41) 344.896 ms 349.207 ms *
12 * * g49.ba01.b002281-1.ord01.atlas.cogentco.com (66.28.66.206) 358.852 ms
13 * * *
14 * 66-28-147-105.servercentral.net (66.28.147.105) 347.723 ms *

i've never seen this line before/this route ....
seems as cogent has bought a feed from sprintlink, which so far is very nice, but it also seems, that this line is so small, maybe just multiple T1's or sth. !? as i started that download file from bassboy too, and here we go:

0K .......... .......... .......... .......... .......... 1% 39:45 1.90K
50K .......... .......... .......... .......... .......... 2% 55:01 1.36K
100K .......... .......... .......... .......... .......... 3% 17:33 4.21K
150K .......... .......... .......... .......... .......... 4% 46:52 1.56K
200K .......... .......... .......... .......... .......... 5% 21:26 3.37K
250K .......... .......... .......... .......... .......... 6% 51:15 1.39K
300K .......... .......... .......... .......... .......... 7% 27:09 2.60K
350K .......... .......... .......... .......... .......... 8% 42:04 1.65K

wtf? it would take HOURS!!!

so it ain't t-offline's fault, it's cogent's fault ...
i think all we can do is wait, and hope it'll get better soon.

greets
Johannes

I am not quite sure if the problem still exists, but link was always slow from Germany to outside world....

Originally posted by apollo
I am not quite sure if the problem still exists, but link was always slow from Germany to outside world....
nope, i had around 80-90kb/s most of the time from some cogent pipes in the US and that's what my pipe can handle.

I did a traceroute too, from Austria, it's fine. 178 ms, no packet loss, no slowness.

Sprint has been one of Cogent's first peering/transit partners. I think they buy connectivity from Above.net and Sprint until they can switch to almost 100% peering.

I'm looking forward to seeing whether they can solve these problems this week-end. So far I've had a good impression Cogent's staff so I'm confident that it will become better by Monday :)

Hello Realist,

no must have been someone else, I haven't been participating on the RS forum yet...

No problems here in Holland. An average ping of 158 and it goes via Frankfurt (Level 3, frankfurt=germany) to verio.

from india


C:\>ping 66.28.0.10

Pinging 66.28.0.10 with 32 bytes of data:

Reply from 66.28.0.10: bytes=32 time=561ms TTL=237
Reply from 66.28.0.10: bytes=32 time=520ms TTL=237
Reply from 66.28.0.10: bytes=32 time=521ms TTL=237
Reply from 66.28.0.10: bytes=32 time=520ms TTL=237

Ping statistics for 66.28.0.10:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 520ms, Maximum = 561ms, Average = 530ms

C:\>ping 66.28.0.10

Pinging 66.28.0.10 with 32 bytes of data:

Reply from 66.28.0.10: bytes=32 time=982ms TTL=237
Reply from 66.28.0.10: bytes=32 time=560ms TTL=237
Reply from 66.28.0.10: bytes=32 time=521ms TTL=237
Reply from 66.28.0.10: bytes=32 time=521ms TTL=237

Ping statistics for 66.28.0.10:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 521ms, Maximum = 982ms, Average = 646ms

C:\>ping 66.28.0.10

Pinging 66.28.0.10 with 32 bytes of data:

Reply from 66.28.0.10: bytes=32 time=571ms TTL=237
Reply from 66.28.0.10: bytes=32 time=510ms TTL=237
Reply from 66.28.0.10: bytes=32 time=531ms TTL=237
Reply from 66.28.0.10: bytes=32 time=551ms TTL=237

Ping statistics for 66.28.0.10:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 510ms, Maximum = 571ms, Average = 540ms
:cool:

Robert,

it's Tuesday and it hasn't changed yet.

I am worried now.

Because I asked in the RS forum and got offered a fix, but it turned out that RS meant to re-route the c-block that I PERSONALLY am surfing on *LOL* - sorry no offense but I wonder what kind of solution that would be.

Like I was concerned about my personal surfing experience. I could always use another ISP but it's not likely for me to convince 10 million german surfers and who knows how many others in other european countries to change ISP before they can access my rackshack servers again

When you said on Friday that this is a PSI /Cogent integration issue and that the connectivity Rackshack->Europe not only should be fixed but probably be even better from Moday on, I was confident, and I would still be, even though it is Tuesday now, but the above mentioned answer in the RS forum frightened me a bit to be honest.

You won't just let it stay that way right?

Again, I am not talking about a personal problem of mine here - I already feel the need to mention this :) - half europe is cut off, I'd say this is a mentionable decrease in Rackshacks overall connectivity,

and one would think that - if cogent/PSI doesn't fix it, which it hasn't yet - Rackshack will be doing something on it's own side, such as a mayor re-arrangement of the outgoing traffic to europe, like re-routing it through another one of your uplinks?

I guess I could come up with the a-blocks of t-online if that would be required for a temporary re-routing.


This is not meant as an offense, and I was happy with rackshack so far. I have 2 servers there and was going to get 2 more this month.

But I would think that Rackshack is going to handle this MAYOR decrease in it's overall connectivity in an appropriate way

I mean, you should be even more concerned about it than I am...?

Maybe I have been misunderstood on the RS forum


Sincerely
Erich

This concerns T-Online and not smaller ISP's in germany, but unfortunately T-Online is the biggest ISP in europe with over 10 million surfers and according to http://www.isp-planet.com/research/2002/usage.html the 6th biggest ISP worldwide.


------

Rackshack Server 64.246.28.78 to ME on ISP "t-online" (very SLOOOOW - connection almost non-existant):

1 64.246.28.1 (64.246.28.1) 0.580 ms 0.702 ms 0.345 ms
2 tayhou-223-2.ev1.net (207.218.223.2) 0.407 ms 0.326 ms 0.303 ms
3 g1.ba21.b000605-0.iah01.atlas.cogentco.com (66.28.28.209) 1.359 ms 1.610 ms 1.205 ms
4 g1-2.core01.iah01.atlas.cogentco.com (66.28.66.233) 1.722 ms 1.499 ms 1.311 ms
5 p15-0.core01.tpa01.atlas.cogentco.com (66.28.4.46) 54.756 ms 223.756 ms 232.380 ms
6 p14-0.core01.jax01.atlas.cogentco.com (66.28.4.142) 26.418 ms 26.075 ms 25.954 ms
7 p5-0.core01.atl01.atlas.cogentco.com (66.28.4.138) 191.645 ms 207.332 ms 203.578 ms
8 154.54.1.2 (154.54.1.2) 31.996 ms 32.009 ms 31.991 ms
9 154.54.1.10 (154.54.1.10) 51.662 ms 52.980 ms 50.462 ms
10 ne.peering.tier1.us.psi.net (154.13.2.34) 54.997 ms 54.914 ms 54.790 ms
11 * * *
12 HH-SB1.HH.DE.net.DTAG.DE (62.156.131.129) 311.481 ms * *
13 L-SB1.L.DE.net.dtag.de (62.154.2.134) 312.505 ms * 312.013 ms
14 * * M-SB1.M.DE.net.dtag.de (62.154.0.234) 311.608 ms
15 M-EB1.M.DE.net.dtag.de (62.154.27.106) 312.334 ms * *
16 217.237.157.93 (217.237.157.93) 313.237 ms 313.719 ms *
17 * MeOnT-Online.t-dialin.net (217.myIP) 375.240 ms *


Fault of PSI/Cogent?


When I am on a small ISP in germany, the connection is o.k.

Rackshack Server to ME on small ISP (not t-online)

30 hops max, 38 byte packets
1 64.246.28.1 (64.246.28.1) 0.689 ms 0.359 ms 0.246 ms
2 tayhou-223-2.ev1.net (207.218.223.2) 1.143 ms 0.566 ms 0.313 ms
3 g1.ba21.b000605-0.iah01.atlas.cogentco.com (66.28.28.209) 1.511 ms 1.431 ms 1.754 ms
4 g1-2.core01.iah01.atlas.cogentco.com (66.28.66.233) 2.069 ms 1.580 ms 1.973 ms
5 p15-0.core01.tpa01.atlas.cogentco.com (66.28.4.46) 21.393 ms 20.981 ms 21.028 ms
6 p14-0.core01.jax01.atlas.cogentco.com (66.28.4.142) 26.243 ms 27.012 ms 25.967 ms
7 p5-0.core01.atl01.atlas.cogentco.com (66.28.4.138) 32.372 ms 33.579 ms 32.286 ms
8 p14-0.core02.dca01.atlas.cogentco.com (66.28.4.126) 44.326 ms 44.366 ms 44.174 ms
9 i00pva-005.ip-plus.net (198.32.190.16) 109.859 ms 50.897 ms 51.206 ms
10 i00nyh-015-pos3-1-0.ip-plus.net (164.128.236.37) 51.036 ms 51.691 ms 50.841 ms
11 i79zhh-005-pos2-2.ip-plus.net (164.128.236.129) 149.494 ms 148.718 ms 149.441 ms
12 i79zhh-015-eth8-0.ip-plus.net (164.128.34.58) 142.652 ms 142.491 ms 142.382 ms
13 i62bsw-000-pos8-1-0.ip-plus.net (164.128.33.86) 143.611 ms 143.569 ms 143.315 ms
14 tesion-bsw.ip-plus.net (164.128.33.62) 148.208 ms 148.087 ms 147.803 ms
15 MeSurfingOnTheSmallISP.ip.tesion.net (195.myIP) 149.392 ms * 149.792 ms


Now ME again on t-online (the bad one) TO -> Rackshack Server:


Routenverfolgung zu 64.246.28.78 über maximal 30 Abschnitte

1 55 ms 55 ms 55 ms 217.myIP
2 55 ms 55 ms 55 ms 217.237.157.90
3 151 ms 151 ms 151 ms Vienna-gw12.USA.net.DTAG.DE [62.156.131.190]
4 151 ms 151 ms 151 ms p4-1-1-0.r00.stngva01.us.bb.verio.net [129.250.9
.197]
5 152 ms 164 ms 151 ms p16-1-0-0.r01.mclnva02.us.bb.verio.net [129.250.
3.157]
6 179 ms 165 ms 165 ms p16-0-0-0.r00.atlnga03.us.bb.verio.net [129.250.
2.49]
7 164 ms 165 ms 179 ms ge-1-2-0.r01.atlnga03.us.bb.verio.net [129.250.5
.246]
8 192 ms 192 ms 192 ms p4-1-3-0.r00.hstntx01.us.bb.verio.net [129.250.2
.45]
9 193 ms 192 ms 178 ms ge-1-1-0.a03.hstntx01.us.ra.verio.net [129.250.3
0.216]
10 192 ms * 192 ms ge-0-0-0.a03.hstntx01.us.ce.verio.net [128.241.2
.102]
11 357 ms * 357 ms tayhou-223-36.ev1.net [207.218.223.36]
12 357 ms * 357 ms

it stops there and/or times out


BTW: Even on t-online I can surf the site http://www.rackshack.net/ just fine. But the Rackshack forum on http://forum.rackshack.net/ is very slow/not available on t-online - just like my Rackshack server.


So it WOULD be possible to have my servers fast too. Just like they were untill Friday.

another trace from t-online:

1 1 ms 1 ms 1 ms *******
2 124 ms 33 ms 137 ms *******
3 34 ms 132 ms 34 ms *******
4 129 ms 129 ms 169 ms Vienna-gw12.USA.net.DTAG.DE [62.156.131.190]
5 126 ms 195 ms 166 ms sl-gw20-rly-2-0.sprintlink.net [160.81.255.9]
6 214 ms 180 ms 174 ms sl-bb23-rly-3-2.sprintlink.net [144.232.14.45]
7 175 ms 127 ms 126 ms sl-bb21-pen-12-0.sprintlink.net [144.232.20.33]
8 127 ms 142 ms 132 ms sl-bb25-pen-9-0.sprintlink.net [144.232.5.238]
9 185 ms 133 ms 172 ms sl-gw4-pen-10-0.sprintlink.net [144.232.5.38]
10 296 ms * * sl-cogentcomm-5-0.sprintlink.net [160.81.202.18]
11 371 ms * * p5-0.core02.jfk02.atlas.cogentco.com [66.28.4.1]
12 * * 378 ms p14-0.core02.ord01.atlas.cogentco.com [66.28.4.86]
13 * * * Zeitüberschreitung der Anforderung.


does seem to have some trouble from t-online.

I only got about 30 KB/s download in the US.

It looks to me like Cogent has reached a point where their transit/peering is filling up. Ex: they have an OC-12 peering session with another ISP and are trying to push OC-12x2 or something. For a more detailed explanation, email me. :)

Cogent -> ATT last night was over 1,024 thanks to the Cogent <-> ATT peering in Chicago being full and/or broken.

This is the glory of $30/Meg bandwidth. This cannot come as a surprise to anyone. I'm sure they'll get it fixed, though. But who knows when it will break again?

Drew

Yessssss

:)

Everything fast again...

keep in mind the large amount of growth that cogent is experiencing, i honestly believe the headsurfer's explanation.

i say go cogent and hopefully they keep growing and growing.

i for one do not like paying out of the ass for bandwidth.