ok... i've got a server in chicago... (and yes, i use cogent only because they're cheap but if there were anybody as easily accessible, i'd drop cogent n run like hell from that cr*p)...

i am on the west coast ... my traceroute to that server takes around 200ms... should not take more than 60ms because not only does my ISP peer with PSINet in chicago (cogent's network after aquisition) but it also peers with metromedia/above.net in seattle... well... i do a traceroute... where does my data go

Vancouver, BC (Telus) -> New Westminister, BC (Telus) -> Seattle, WA (Telus) -> Seattle, WA (MetroMedia/Above.net) -> San Jose, CA (MetroMedia/Above.net) -> Pao Alto, CA (MetroMedia/Above.net) -> San Francisco, CA (Cogent) -> Dallas/Fort Worth, TX (Cogent) -> Washington, DC (Cogent) -> New York, NY (Cogent) -> Chicago, IL (Cogent)

i mean... whats the point of going bloody coast-to-coast on their own network... i could understand if the entry point from other networks made it go coast to coast... but ... this is bloody ridiculous... any cogent lover want to justify this? oh... and 200ms, that does make a bloody difference when loading websites! i mean... ping times like that... make even my SSH and Terminal Services sessions bloody unbearable!

not to forget i get minimum 80ms pings to stuff on the west coast when it shouldn't be more than 30-40! i mean... i can get to my exodus stuff in Georgia in around 80ms!

You have no right to complain! Cogent is ~30$ per mbit. Yipes, Time Warner etc providers are ~300$ per mbit. Yes that's right! 10 times more! Tier 1 providers cost even more.

Cogent offers you cheap bandwith, so it is no issue on HOW they do it cause with that price you can be glad if it works at all (and it does!)

nice argument... $30/Mbps but thats a minimum commitment of 100Mbps... verio is no more than $90/Mbps on most major providers... and thats in low volumes... yet their bandwidth is probably what, 10x better than cogent? so shouldn't cogent reduce their bandwidth rate to $9/Mbps then? same with level3/etc, you get 100Mbps, then ~$100/Mbps... so cogent should be $10-9/Mbps... on a commitment of 100Mbps...

even with that cheap bullsh*t, there should atleast be some level of quality... ! paying for this crap gives me the right to complain! just because i buy a ford doesn't mean i can't b*tch at the sales person if my steering wheel breaks off... ! or the car stalls at anything above 30mph... !

heck dude... i could sell a 100Mbps pipe with the full 100Mbps and rack for $9000/month with exodus+internap+worldcom bandwidth colocated directly in an exodus facility... and thats WITHOUT overselling... just because its cheap doesn't mean i am not responsible for the quality of service! thats only what, $90/Mbps? and those are the top 3 tier-1 providers (well, excluding internap i guess)...

I don't see any problem with that route.

I've never seen a route where it went from Seattle(Above.net) to Cogent. My traces always go to Palo Alto, where Cogent peers with Above.net.

Besides, even Cogent's network map, shows that they go from Seattle, to Sacremento. Nowhere does it show Seattle going anywhere else.

the problem with my route is that it goes from west coast to east coast to central... a totally pointless addition of latency... all within the cogent network

Who says that Cogent HAS to be 10 times cheaper. Their service is much more then just 10% of Verio and others.

Even if Cogent was "just" 50% cheaper it would still be a good offer cause it would still be half the price... anyway I really don't see an issue here.

Originally posted by netdude
the problem with my route is that it goes from west coast to east coast to central... a totally pointless addition of latency... all within the cogent network


The longer your packets stay on their network, the less it costs cogent...

Thats how they offer low cost bandwidth..


If you want better peering, then pay more for better bandwidth, plain and simple....

but i guess i could say... pretty much all the other major backbones are atleast 3x better than cogent... 60ms to chicago vs 200ms... heh

nocsol, i see you're in chicago as well... heh... i do a traceroute to chicago.com (hosted by exodus in chicago)... i get 60ms, 13 hops... ... ... why does cogent take 200ms and 23 hops? because they're so excellent ... ... ...? and no, my isp does not peer with exodus

Originally posted by netdude
but i guess i could say... pretty much all the other major backbones are atleast 3x better than cogent... 60ms to chicago vs 200ms... heh

nocsol, i see you're in chicago as well... heh... i do a traceroute to chicago.com (hosted by exodus in chicago)... i get 60ms... ... ... why does cogent take 200ms? because they're so excellent ... ... ...?

No mate. Because they are 10 times cheaper then all the others... geeze :rolleyes:

incorrect... i am sure i could get two routers, put it in the exodus datacenter and get them to bounce every packet between eachother atleast a few hundred times before putting it out over the exodus feed, but i doubt that'll save me a penny... their cost is dependant on trying to put the packets over their peers rather than their transit provider (above.net)... well, once data is already on the cogent network, theres no point in routing their data back n forth, coast to coast... its a waste of network resources actually...

Originally posted by The Prohacker



The longer your packets stay on their network, the less it costs cogent...

Thats how they offer low cost bandwidth..


If you want better peering, then pay more for better bandwidth, plain and simple....

they aren't 10 times cheaper... 3x cheaper than my example offer, which includes rackspace, so cogent ain't all that great overall... and my offer includes bandwidth from those tier-1s that u say cost so much more...

and i already gave you examples of 2 of those alternative providers (verio and level3) that you mentioned... and how cogent definately isn't a tenth of their price...

Originally posted by NocSol


No mate. Because they are 10 times cheaper then all the others... geeze :rolleyes:

Yeah well, that might be correct. My point is more of a non-technial nature. I pay 10% of what others ask (or 20% or 30%, doesn't really matter) and I have ~98% uptime. An dwhen the routing sometimes sucks (300ms+) then that's ok too cause it comes with the low charge. If they had no disadvantage at that pricing, then all other bw-providers would b eout of business.

If you aren't willing to pay more, then don't compain.

I'm north of Seattle, WA and I just did a PING to Chicago(fdcservers.net) and it was only 110ms.

Originally posted by dreamHOBO
If you aren't willing to pay more, then don't compain.

EXACTLY.

Originally posted by JTY
I'm north of Seattle, WA and I just did a PING to Chicago(fdcservers.net) and it was only 110ms

Yep, most of the time it is really good... and even when it's "bad" it still works ;)

This is a traceroute to kingcomp.net (the only ISP using Cogent in Chicago I can think of at the moment )

traceroute to kingcomp.net (66.28.242.138), 30 hops max, 40 byte packets
1 mfsc1-130-150-102.csu.net (130.150.102.253) 0.354 ms 0.227 ms 0.222 ms
2 WestEdGSRPOS-WestEdCAT6009POS.CSU.net (137.145.11.45) 0.239 ms 0.210 ms 0.206 ms
3 QSVGSR-WEGSR-ATM.CSU.net (137.145.202.125) 11.023 ms 10.943 ms 10.864 ms
4 QSV-M10-C2.GE.calren2.net (137.164.12.166) 20.994 ms 11.348 ms 11.172 ms
5 POS1-0.hsipaccess1.SanJose1.Level3.net (209.247.159.109) 11.621 ms 11.548 ms 11.693 ms
6 gigabitethernet4-2.core1.SanJose1.Level3.net (64.159.2.133) 11.646 ms 11.545 ms 11.586 ms
7 pos5-0.mpr6.sjc2.us.mfnx.net (208.184.232.161) 11.375 ms 11.906 ms 11.443 ms
8 pos0-0.mpr2.sjc2.us.mfnx.net (208.185.156.69) 11.736 ms 11.412 ms 11.411 ms
9 pos9-0.mpr1.sjc2.us.mfnx.net (216.200.0.233) 11.999 ms 11.455 ms 11.670 ms
10 so-1-1-0.mpr4.sjc2.us.mfnx.net (208.184.102.202) 11.544 ms 11.754 ms 11.548 ms
11 pos7-0.mpr1.iad1.us.mfnx.net (216.200.127.25) 78.954 ms 79.035 ms 79.606 ms
12 so-2-0-0.cr1.iad1.us.mfnx.net (208.185.0.138) 79.275 ms 79.484 ms 79.174 ms
13 pos2-0.mpr1.iad5.us.mfnx.net (216.200.127.10) 79.474 ms 79.343 ms 79.118 ms
14 64.124.112.29.cogentco.com (64.124.112.29) 89.992 ms 89.661 ms 90.547 ms
15 p15-0.core02.dca01.atlas.cogentco.com (66.28.4.22) 90.221 ms 90.985 ms 89.962 ms
16 p6-0.core01.jfk02.atlas.cogentco.com (66.28.4.82) 86.681 ms 85.876 ms 85.716 ms
17 p15-0.core02.jfk02.atlas.cogentco.com (66.28.4.14) 186.117 ms 205.000 ms 200.751 ms
18 p14-0.core02.ord01.atlas.cogentco.com (66.28.4.86) 86.092 ms 86.650 ms 86.438 ms
19 g50.ba02.b000268-0.ord01.atlas.cogentco.com (66.28.66.78) 86.465 ms 85.776 ms 87.006 ms
20 Datacenter.demarc.cogentco.com (66.28.21.82) 92.227 ms 94.982 ms 87.212 ms
21 66.28.242.138 (66.28.242.138) 88.530 ms 88.021 ms 92.167 ms


That's 92ms maximum.

Now, a traceroute from the same point to chicago.com:

traceroute to chicago.com (64.37.221.182), 30 hops max, 40 byte packets
1 mfsc1-130-150-102.csu.net (130.150.102.253) 0.398 ms 0.230 ms 0.239 ms
2 WestEdGSRPOS-WestEdCAT6009POS.CSU.net (137.145.11.45) 0.253 ms 0.217 ms 0.211 ms
3 QSVGSR-WEGSR-ATM.CSU.net (137.145.202.125) 11.034 ms 10.939 ms 10.945 ms
4 QSV-M10-C2.GE.calren2.net (137.164.12.166) 11.476 ms 11.289 ms 11.030 ms
5 POS1-0.hsipaccess1.SanJose1.Level3.net (209.247.159.109) 11.756 ms 11.498 ms 11.520 ms
6 ae0-56.mp2.SanJose1.Level3.net (64.159.2.161) 11.855 ms 11.981 ms 11.940 ms
7 so-2-0-0.mp2.SanFrancisco1.Level3.net (64.159.0.217) 12.880 ms 12.742 ms 12.893 ms
8 pos9-0.core1.SanFrancisco1.Level3.net (209.247.10.234) 12.903 ms 13.152 ms 12.748 ms
9 acr1-so-2-0-0-0.SantaClara.cw.net (208.172.147.201) 14.779 ms 14.717 ms 14.528 ms
10 agr4-loopback.SantaClara.cw.net (208.172.146.104) 15.045 ms 15.095 ms 15.219 ms
11 dcr1-so-1-3-0.SantaClara.cw.net (208.172.156.29) 14.942 ms 14.809 ms 14.743 ms
12 ibr01-p5-0.sntc05.exodus.net (208.172.156.74) 15.634 ms 15.915 ms 15.782 ms
13 bbr01-g6-0.sntc05.exodus.net (64.56.192.17) 16.010 ms 16.093 ms 16.025 ms
14 bbr02-p2-0.sntc03.exodus.net (209.1.169.69) 17.055 ms 17.263 ms 17.153 ms
15 bbr02-p0-0.okbr01.exodus.net (216.32.132.170) 77.594 ms 77.889 ms 78.030 ms
16 dcr03-g8-0.okbr01.exodus.net (216.34.183.100) 77.620 ms 77.473 ms 77.486 ms
17 csr02-ve242.okbr01.exodus.net (216.33.64.18) 78.300 ms 78.332 ms 78.121 ms
18 64.37.221.182 (64.37.221.182) 77.863 ms 77.736 ms 77.819 ms

That's 77 ms, that's a difference of 15ms.

I don't get your point. I can give you examples of bad routing for every provider out there. Packets often take ridiculous routes all over the globe. That's not really unique to Cogent.

If you post the actual traceroute, we can determine where most of the latency is coming from.

Originally posted by JTY
I'm north of Seattle, WA and I just did a PING to Chicago(fdcservers.net) and it was only 110ms.
The ping from the UK:

Pinging fdcservers.net [66.28.242.138] with 32 bytes of data:

Reply from 66.28.242.138: bytes=32 time=128ms TTL=49
Reply from 66.28.242.138: bytes=32 time=126ms TTL=49
Reply from 66.28.242.138: bytes=32 time=130ms TTL=49
Reply from 66.28.242.138: bytes=32 time=129ms TTL=49

Speed is really good :) (much, much better than what I normally get for some hosts who are on a different provider other than Cogent), I wouldn’t complain if my host had pings like that!

so... nocsol, when your clients ask you what kind of service they should expect from you, what do you tell them? 'you get what you pay for'... kinda like what my friend says that works at a chevron says... "they pretend to pay me, so i pretend to work'... lol

also... u got a GigE to cogent in chicago? i didn't even think fdc had that... lol...

Originally posted by NocSol
Yeah well, that might be correct. My point is more of a non-technial nature. I pay 10% of what others ask (or 20% or 30%, doesn't really matter) and I have ~98% uptime. An dwhen the routing sometimes sucks (300ms+) then that's ok too cause it comes with the low charge. If they had no disadvantage at that pricing, then all other bw-providers would b eout of business.

the route just changed recently for me... but still sh*tty

Tracing route to www.kingcomp.net [66.28.242.138]
over a maximum of 30 hops:

1 20 ms 10 ms 20 ms 209.53.12.254
2 10 ms 10 ms 50 ms 209.53.75.198
3 20 ms 20 ms 20 ms nwmrbc01br01.bb.telus.com [204.174.217.39]
4 40 ms 20 ms 10 ms sttlwa01br02.bb.telus.com [209.53.75.178]
5 20 ms 20 ms 20 ms acr2-so-2-1-0-0.Seattle.cw.net [208.172.81.129]
6 40 ms 30 ms 40 ms agr4-loopback.SantaClara.cw.net [208.172.146.104]
7 30 ms 30 ms 30 ms dcr1-so-1-3-0.SantaClara.cw.net [208.172.156.29]
8 40 ms 40 ms 41 ms agr2-so-0-0-0.SantaClara.cw.net [208.172.156.58]
9 30 ms 30 ms 40 ms bpr1-loopback.PaloAltoPaix.cw.net [208.172.146.18]
10 30 ms 40 ms 30 ms 206.24.241.42
11 40 ms 30 ms 40 ms so-4-1-0.mpr4.sjc2.us.mfnx.net [208.185.175.161]
12 91 ms 100 ms 90 ms pos7-0.mpr1.iad1.us.mfnx.net [216.200.127.25]
13 90 ms 100 ms 91 ms so-2-0-0.cr1.iad1.us.mfnx.net [208.185.0.138]
14 90 ms 100 ms 90 ms pos2-0.mpr1.iad5.us.mfnx.net [216.200.127.10]
15 * 150 ms 150 ms 64.124.112.29.cogentco.com [64.124.112.29]
16 150 ms 151 ms 140 ms p15-0.core02.dca01.atlas.cogentco.com [66.28.4.22]
17 150 ms 161 ms 150 ms p6-0.core01.jfk02.atlas.cogentco.com [66.28.4.82]
18 160 ms 150 ms 161 ms p15-0.core02.jfk02.atlas.cogentco.com [66.28.4.14]
19 180 ms 180 ms 181 ms p14-0.core02.ord01.atlas.cogentco.com [66.28.4.86]
20 180 ms 180 ms 181 ms g50.ba02.b000268-0.ord01.atlas.cogentco.com [66.28.66.78]
21 * 181 ms 180 ms Datacenter.demarc.cogentco.com [66.28.21.82]
22 180 ms 181 ms 180 ms 66.28.242.138

Trace complete.

No reason to attack me mate. The GigE is NOT in our Chicago facility nor did I ever say it was.

When my clients ask me I show them our 98% SLA that applies to Cogent in Chicago and of course they are aware that it is Cogent. Again I don't really see an issue here nor why you are mixing my business into your discourse with Cogent...

Trace is pretty good from here. If Kingcomps is yours, you should really upgrade apache.


------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Hop | %Loss | IP Address | Node Name | Location | Tzone | ms | Graph | Network |
------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | | xx.xx.xxx.xx | **** | * | | | | ServiceCo LLC - Road Runner |
| 1 | | 10.47.96.1 | - | ... | | 42 | -x- | (private use) |
| 2 | | 24.29.100.41 | 24-29-100-41.nyc.rr.com | New York, NY, USA | -05:00 | 15 | x-- | ServiceCo LLC - Road Runner |
| 3 | | 24.29.100.90 | pos2-0-nycmnyr-rtr1.nyc.rr.com | New York, NY, USA | -05:00 | 17 | x- | ServiceCo LLC - Road Runner |
| 4 | 10 | 24.29.97.29 | pos3-0-nycmnya-rtr1.nyc.rr.com | New York, NY, USA | -05:00 | 10 | x- | ServiceCo LLC - Road Runner |
| 5 | | 66.185.137.193 | pop1-nye-P0-2.atdn.net | ?Reston, VA 20191 | | 9 | x-- | AOL Transit Data Network |
| 6 | | 66.185.141.18 | bb2-nye-P5-0.atdn.net | ?Reston, VA 20191 | | 22 | x- | AOL Transit Data Network |
| 7 | | 66.185.153.118 | bb2-vie-P8-0.atdn.net | ?Reston, VA 20191 | | 47 | -x-------- | AOL Transit Data Network |
| 8 | | 66.185.139.83 | pop1-vie-P5-0.atdn.net | ?Reston, VA 20191 | | 48 | -x-- | AOL Transit Data Network |
| 9 | | 209.249.203.229 | pos8-0.er1a.iad1.us.mfnx.net | ?San Jose, CA 95113 | | 10 | x | Abovenet Communications, Inc. |
| 10 | | 208.185.0.153 | so-3-2-0.cr1.iad1.us.mfnx.net | ?San Jose, CA 95113 | | 15 | x- | Abovenet Communications, Inc. |
| 11 | | 216.200.127.10 | pos2-0.mpr1.iad5.us.mfnx.net | ?San Jose, CA 95113 | | 12 | x- | Abovenet Communications, Inc. |
| 12 | | 64.124.112.29 | 64.124.112.29.cogentco.com | - | | 12 | x- | Abovenet Communications, Inc. |
| 13 | | 66.28.4.22 | p15-0.core02.dca01.atlas.cogentco.com | Washington, DC, USA | -05:00 | 26 | x------ | Cogent Communications |
| 14 | | 66.28.4.82 | p6-0.core01.jfk02.atlas.cogentco.com | New York, NY, USA | -05:00 | 21 | x-- | Cogent Communications |
| 15 | | 66.28.4.14 | p15-0.core02.jfk02.atlas.cogentco.com | New York, NY, USA | -05:00 | 31 | -x----- | Cogent Communications |
| 16 | | 66.28.4.86 | p14-0.core02.ord01.atlas.cogentco.com | Chicago, IL, USA | -06:00 | 47 | x-- | Cogent Communications |
| 17 | | 66.28.66.78 | g50.ba02.b000268-0.ord01.atlas.cogentco.com | Chicago, IL, USA | -06:00 | 46 | x- | Cogent Communications |
| 18 | | 66.28.21.82 | Datacenter.demarc.cogentco.com | - | | 48 | x- | Cogent Communications |
| 19 | | 66.28.242.138 | www.kingcomp.net | ?Washington, DC 20007 | | 46 | x-- | Cogent Communications |
------------------------------------------------------------------------------------------------------------------------------------------------------------------
Roundtrip time to www.kingcomp.net, average = 46ms, min = 40ms, max = 110ms -- 09-Aug-02 7:32:31 PM

nocsol... sorry dude (i assumed, incorrectly) ... i shutup now ... because its obvious u guys have proven ur point ... and i've kicked and screamed enough ... lol

Originally posted by Gernot
I don't get your point. I can give you examples of bad routing for every provider out there. Packets often take ridiculous routes all over the globe. That's not really unique to Cogent.

gernot, i'd love to see you find one for exodus or internap...

Originally posted by netdude
nocsol... sorry dude (i assumed, incorrectly) ... i shutup now ... because its obvious u guys have proven ur point ... and i've kicked and screamed enough ... lol

I would have aprechiated a professional discussion rather then "kicked and screamed" actually :(

With pleasure.

traceroute to chicago.com (64.37.221.182), 30 hops max, 40 byte packets
1 phnx-core-7513.getnet.net (216.19.223.1) 1.113 ms 0.523 ms 0.441 ms
2 host9.64-58-236.webaccess.net (64.58.236.9) 1.029 ms 1.263 ms 1.014 ms
3 gige5-0-144.hsipaccess2.Phoenix1.Level3.net (63.214.168.41) 1.255 ms 1.029 ms 1.208 ms
4 ge-6-0-1.mp1.Phoenix1.Level3.net (64.159.3.105) 1.673 ms 2.054 ms 1.561 ms
5 so-3-0-0.mp2.SanJose1.Level3.net (64.159.1.130) 18.630 ms 19.142 ms 18.918 ms
6 so-2-0-0.mp2.SanFrancisco1.Level3.net (64.159.0.217) 19.763 ms 19.076 ms 18.971 ms
7 pos9-0.core1.SanFrancisco1.Level3.net (209.247.10.234) 19.501 ms 18.952 ms 20.165 ms
8 acr1-so-2-0-0-0.SantaClara.cw.net (208.172.147.201) 20.946 ms 22.074 ms 21.011 ms
9 agr4-loopback.SantaClara.cw.net (208.172.146.104) 22.899 ms 21.379 ms 21.232 ms
10 dcr1-so-0-3-0.SantaClara.cw.net (208.172.156.13) 21.503 ms 21.023 ms 21.225 ms
11 ibr01-p5-0.sntc05.exodus.net (208.172.156.74) 22.348 ms 22.404 ms 22.251 ms
12 bbr01-g2-0.sntc05.exodus.net (64.56.192.1) 22.136 ms 22.068 ms 22.754 ms
13 bbr02-p2-0.sntc03.exodus.net (209.1.169.69) 23.856 ms 23.120 ms 24.402 ms
14 bbr02-p0-0.okbr01.exodus.net (216.32.132.170) 77.998 ms 77.023 ms 78.122 ms
15 dcr03-g7-0.okbr01.exodus.net (216.34.183.68) 77.613 ms 77.392 ms 77.887 ms
16 csr02-ve242.okbr01.exodus.net (216.33.64.18) 78.314 ms 77.951 ms 77.866 ms
17 64.37.221.182 (64.37.221.182) 77.641 ms 77.511 ms 77.932 ms


Why does it go from Phoenix to the west coast and not directly to Chicago? That's highly inefficient. If I had more time I would love to give you more examples but unfortunately, I have other things to do as well :)

Routing is a little more complicated than you think, so you'll always see such inefficient routes with every single carrier. That's a problem of the Internet as a whole and not solely of Cogent.

Originally posted by Gernot
With pleasure.

Why does it go from Phoenix to the west coast and not directly to Chicago? That's highly inefficient. If I had more time I would love to give you more examples but unfortunately, I have other things to do as well :)

Routing is a little more complicated than you think, so you'll always see such inefficient routes with every single carrier. That's a problem of the Internet as a whole and not solely of Cogent.

why? because thats level3's fault... not exodus' ... level3 passes the data onto exodus in santa clara... up until that point, exodus has no clue where the data is coming from... exodus' route is definately most efficient... adding only ~55ms from west coast to central... !

my original complaint was why cogent was taking the data from san francisco, piping to to new york, then washington dc and then to chicago... essentially making it bounce from coast to coast... within the cogent network...

I'll just give you one more example of a great route:

1 vlan110.r-s-fcc2-server.fnal.gov 131.225.110.200 0.567 1.942 0.302
2 ge5-0-0.r-s-frw.fnal.gov 131.225.15.5 0.351 0.308 0.312
3 r-x-esnet.fnal.gov 198.49.208.4 0.624 0.439 0.591
4 chi-fnal.es.net 134.55.208.17 3.345 2.029 1.877
5 snv-s-chi.es.net 134.55.205.101 50.297 50.367 56.589
6 paix-pa-snv.es.net 134.55.208.205 51.313 53.260 51.099
7 Not Registered 206.24.240.201 51.109 51.156 51.085
8 ibr01-p1-0.paix01.exodus.net 216.32.132.157 50.829 52.697 50.915
9 ibr02-g1-0.paix01.exodus.net 206.79.9.242 53.458 51.267 54.558
10 bbr01-p6-0.sntc03.exodus.net 209.185.9.241 52.054 54.256 53.682
11 bbr02-g4-1.sntc03.exodus.net 216.33.153.82 57.774 55.051 54.506
12 bbr02-p0-0.okbr01.exodus.net 216.32.132.170 121.898 120.823 121.140
13 dcr03-g7-0.okbr01.exodus.net 216.34.183.68 124.329 120.900 121.381
14 csr02-ve242.okbr01.exodus.net 216.33.64.18 121.017 121.474 121.544
15 Not Registered 64.37.221.182 120.761 120.857 138.176

It goes from Chicago to the west coast (PAIX) and then goes back to Chicago.

Guess I'll have to open a thread with the title Why Exodus is bullsh*t

Originally posted by netdude


why? because thats level3's fault... not exodus' ... level3 passes the data onto exodus in santa clara... up until that point, exodus has no clue where the data is coming from... exodus' route is definately most efficient... adding only ~55ms from west coast to central... !

Ok, bu thow much does exodus or similar providers cost per mbit? ;)

gernot, there ya go again... exodus' route is perfectly efficient...! no flaws whatsoever... its the preceeding provider that decides to hand off data at a useless point... (probably to take advantage of inexpensive peering rather than putting it over transit in chicago)

and nocsol... i told u... i'd happily sell you exodus bandwidth at $90/Mbps on a commitment of 100Mbps... (including a rack in an exodus facility too!)... and additional bandwidth at $30-60/Mbps depending on volume, up to GigE over fiber... the price isn't always the issue... ! anybody with even moderate sales skills can get quotes like this... from any major provider... its not hard... these quotes, i have gotten, in writing!

Originally posted by netdude
and nocsol... i told u... i'd happily sell you exodus bandwidth at $90/Mbps on a commitment of 100Mbps... (including a rack in an exodus facility too!)... and additional bandwidth at $30-60/Mbps depending on volume, up to GigE over fiber... the price isn't always the issue... ! anybody with even moderate sales skills can get quotes like this... from any major provider... its not hard... these quotes, i have gotten, in writing!

Ok I see your point but you need to see mine too. Cogent is still 3 times cheaper. 3 times is A LOT!

Originally posted by netdude

why? because thats level3's fault... not exodus' ... level3 passes the data onto exodus in santa clara... up until that point, exodus has no clue where the data is coming from... exodus' route is definately most efficient... adding only ~55ms from west coast to central... !

my original complaint was why cogent was taking the data from san francisco, piping to to new york, then washington dc and then to chicago... essentially making it bounce from coast to coast... within the cogent network...

You said:

Vancouver, BC (Telus) -> New Westminister, BC (Telus) -> Seattle, WA (Telus) -> Seattle, WA (MetroMedia/Above.net) -> San Jose, CA (MetroMedia/Above.net) -> Pao Alto, CA (MetroMedia/Above.net) -> San Francisco, CA (Cogent) -> Dallas/Fort Worth, TX (Cogent) -> Washington, DC (Cogent) -> New York, NY (Cogent) -> Chicago, IL (Cogent)


From SF, to Dallas to Washington, DC to New York and then up to Chicago. That's not really bouncing from coast to coast.

gernot, you're right... its going from one coast all the way to the other... and then to the center... when it would have been just easier to go to the center in the first place...

cuz i know there are feeds between dallas and chicago on the cogent network... i've seen 'em myself... because i had this messed up routing before where my data would go to chicago on my ISP, then to metromedia in chicago... and then to dallas on metromedia... then cogent (in dallas) would take that either to florida and then to chicago or back to chicago directly...

i have no doubt the cogent network itself is decent... its their routing that needs a major overhaul... they did that, they could actually compete ... but they can't do that because nobody wants to peer with them... lol

Originally posted by netdude
gernot, there ya go again... exodus' route is perfectly efficient...! no flaws whatsoever... its the preceeding provider that decides to hand off data at a useless point... (probably to take advantage of inexpensive peering rather than putting it over transit in chicago)


That's exactly my point as well. Cogent's route is also perfectly efficient, considering that unlike Exodus and other carriers (having been in Chapter 11) they haven't layed fiber between each city.

The only inefficiency I can see with your example is that it could have taken the direct route from Washington, DC to Chicago. But for that, they would need direct fiber between these two cities which would be the kind of non-sense all those Tier 1 carriers used to do. Such projects, while costing millions of dollars, are completely useless.

these other carriers have routes between most major cities already... heh... if cogent were smart and just bought up dark fiber instead of trying to be cool and lay some new fiber for a DWDM network (which could have saved a boatload too) and bought better routes between major cities...

don't forget, cogent isn't profitable... its a matter of time that if they don't turn a profit, they'll b in chapter 11 too... and nobody wants to really buy them up, because if they did, they would have to raise prices in order to not alienate their current clients... but if they did, everybody would drop them...

as for cogent's route being efficient... is everything i've said going in one ear and out the other? cogent's route in my case = san fran -> dallas -> new york -> dc -> chicago

when i know they have a route specifically: san fran -> dallas -> chicago

that save a lot of latency!

Look, such a traceroute:

traceroute to kingcomp.net (66.28.242.138), 30 hops max, 38 byte packets
1 f13.ba01.b003070-1.sfo01.atlas.cogentco.com (66.250.4.165) 2.860 ms 0.530 ms 6.181 ms
2 g0-2.core02.sfo01.atlas.cogentco.com (66.28.5.181) 0.876 ms 0.586 ms 0.906 ms
3 p15-0.core01.sfo01.atlas.cogentco.com (66.28.4.69) 50.328 ms 3.018 ms 0.658 ms
4 p14-0.core01.ord01.atlas.cogentco.com (66.28.4.41) 46.458 ms 46.425 ms 46.524 ms
5 g49.ba02.b000268-0.ord01.atlas.cogentco.com (66.28.66.74) 46.891 ms 46.546 ms 46.714 ms
6 Datacenter.demarc.cogentco.com (66.28.21.82) 48.757 ms 47.724 ms 53.068 ms
7 66.28.242.138 (66.28.242.138) 53.260 ms 56.365 ms 51.221 ms

doesn't mean that there's a direct route between San Francisco and Chicago. It still goes through Dallas or Denver or wherever. There's also no direct route between Dallas and Chicago, just because you only see these two cities on the traceroute. It still goes through a router in Kansas City which is just not shown on the traceroute.

Enough for today though :)

Yeah I am off 2 bed too... 3:30am here *yawn*

gernot... that route u just pasted... now why can't cogent just provide that when i do a trace from here? cuz that'd make life a lot easier... and make their network look sooo much faster... so its kinda dumb for them not to... heh

morons working in their network design dept?

i mean... thats san fran directly to chicago... that just b a sweet route... heh... latency is even excellent on it... heh

why does it not show the hops in between if there are routers in between... why does it show hops when i do a trace to say dc... ? what u say doesn't make sense... i know what ur getting at, but it don't make sense... plus... the latency is a bit too low for there to b other points in between...

Originally posted by Gernot
From SF, to Dallas to Washington, DC to New York and then up to Chicago. That's not really bouncing from coast to coast.

It's more like following the US coast in a counter-clockwise fashion. I tried visualizing this, and got a sort of swirly thingy.

I'm not sure which is worse -- bouncing between coasts, or going in large swirls. But either way, it is clearly flawed. :bawling:

But yeah, every single complaint I've read about Cogent (well, every reasonable one... I'm excluding the occasional lunatic rants of people who haven't used Cogent) is about how the routing adds enormous latency. I've never used Cogent (so I'll try to refrain from "lunatic rants" :cool: ), but the network route does seem pretty messed up.

Course, people have actually seen employees of my ISP smoking pot while at work, so I should expect some REALLY weird routing here in comparison to Cogent. :eek:

I think this kinda shows you get what you pay for.


When you pay $30/mbit

[root@ns1 root]# traceroute www.fdcservers.net
traceroute to fdcservers.net (66.28.242.138), 30 hops max, 38 byte packets
1 63.239.59.1 (63.239.59.1) 2.016 ms 2.830 ms 1.946 ms
2 chcg01-edge19.il.inet.qwest.net (63.237.33.33) 9.808 ms 9.944 ms 10.712 ms
3 chcg01-core03.il.inet.qwest.net (205.171.20.189) 10.181 ms 10.296 ms 10.112 ms
4 chcg01-core01.il.inet.qwest.net (205.171.20.173) 11.091 ms 11.735 ms 13.579 ms
5 nwrk01-core02.nj.inet.qwest.net (205.171.8.230) 31.682 ms 31.461 ms 31.474 ms
6 nwrk01-core03.nj.inet.qwest.net (205.171.17.34) 31.566 ms 31.464 ms 31.485 ms
7 wash02-core02.dc.inet.qwest.net (205.171.8.181) 34.333 ms 34.320 ms 34.257 ms
8 wash02-brdr01.dc.inet.qwest.net (205.171.9.54) 34.607 ms 34.286 ms 34.218 ms
9 pos8-0.per1.iad1.us.mfnx.net (208.184.233.33) 34.638 ms 34.885 ms 34.736 ms
10 so-3-0-0.cr1.iad1.us.mfnx.net (208.185.0.142) 34.990 ms 34.907 ms 34.905 ms
11 pos2-0.mpr1.iad5.us.mfnx.net (216.200.127.10) 41.704 ms 34.760 ms 34.796 ms
12 64.124.112.29.cogentco.com (64.124.112.29) 48.604 ms 48.352 ms 48.360 ms
13 p15-0.core02.dca01.atlas.cogentco.com (66.28.4.22) 48.869 ms 48.803 ms 48.762 ms
14 p6-0.core01.jfk02.atlas.cogentco.com (66.28.4.82) 54.635 ms 54.537 ms 54.478 ms
15 p15-0.core02.jfk02.atlas.cogentco.com (66.28.4.14) 58.079 ms 55.165 ms 54.879 ms
16 p14-0.core02.ord01.atlas.cogentco.com (66.28.4.86) 61.145 ms 60.872 ms 61.309 ms
17 g50.ba02.b000268-0.ord01.atlas.cogentco.com (66.28.66.78) 61.196 ms 61.142 ms 61.214 ms
18 Datacenter.demarc.cogentco.com (66.28.21.82) 71.085 ms 64.363 ms 69.520 ms
19 66.28.242.138 (66.28.242.138) 68.553 ms 68.553 ms 68.730 ms


When you pay $300-400/mbit

(unless your hostmania)
[root@ns1 root]# traceroute www.anet.com
traceroute to www.anet.com (207.230.26.171), 30 hops max, 38 byte packets
1 63.239.59.1 (63.239.59.1) 2.336 ms 1.827 ms 1.564 ms
2 t38098-1.uschcg2.bsn.savvis.net (209.44.21.5) 14.841 ms 12.772 ms 12.620 ms
3 at-1-2-0.702.uschcg-j20c.savvis.net (64.241.88.41) 12.359 ms 12.778 ms 12.525 ms
4 500.POS2-1.GW6.CHI6.ALTER.NET (157.130.116.201) 14.459 ms 12.627 ms 12.386 ms
5 internap3-gw.customer.alter.net (157.130.121.14) 11.552 ms 11.902 ms 11.684 ms
6 border3.fe1-0-bbnet2.chi.pnap.net (216.52.128.71) 11.475 ms 11.696 ms 11.626 ms
7 anet-1.border3.chi.pnap.net (216.52.130.98) 92.168 ms 154.338 ms 235.509 ms
8 core1.nap-1.il.anet.com (198.77.86.6) 13.122 ms 13.114 ms 13.474 ms
9 207.230.26.171 (207.230.26.171) 13.886 ms 13.051 ms 13.316 ms
[root@ns1 root]#

[root@ns1 root]# traceroute www.chicago.com
traceroute to www.chicago.com (64.37.221.182), 30 hops max, 38 byte packets
1 63.239.59.1 (63.239.59.1) 2.032 ms 1.797 ms 2.352 ms
2 chcg01-edge19.il.inet.qwest.net (63.237.33.33) 9.833 ms 9.948 ms 9.858 ms
3 chcg01-core02.il.inet.qwest.net (205.171.20.65) 10.153 ms 10.355 ms 10.216 ms
4 chcg08-brdr01.il.inet.qwest.net (205.171.220.58) 10.560 ms 10.560 ms 10.517 ms
5 bpr1-so-5-0-0-0.ChicagoEquinix.cw.net (208.174.226.29) 10.553 ms 10.518 ms 10.621 ms
6 agr1-loopback.Chicago.cw.net (208.172.2.101) 49.367 ms 10.950 ms 10.878 ms
7 dcr1-so-6-0-0.Chicago.cw.net (208.175.10.49) 10.983 ms 10.753 ms 10.618 ms
8 ibr02-p6-0.okbr01.exodus.net (208.175.10.82) 11.636 ms 11.789 ms 11.498 ms
9 dcr03-g8-0.okbr01.exodus.net (216.34.183.100) 11.741 ms 11.496 ms 11.622 ms
10 csr02-ve242.okbr01.exodus.net (216.33.64.18) 12.402 ms 12.482 ms 12.152 ms
11 64.37.221.182 (64.37.221.182) 12.007 ms 11.948 ms 11.981 ms
[root@ns1 root]#


Now either quit your whining or go buy from a non-cogent provider.

but i demand to whine... my bloody bank taking so long to approve my financing... ERR... once they do, i kiss cogent good bye forever... lol... :)

*sit in corner and pout 'till I get money*

i just noticed something... is it just me or did everybody drop their peers with cogent?????? everything seems to be going through metromedia... and if so, how much longer can cogent live with transit rates?

I think I understand your point netdude. I hate when people here always answer with the worn-out cliche "you get what you pay for". It just is not always applicable.

exactly... i mean... my only response is... if i buy a brand new dodge neon and the steering wheel comes off... what do i do? just accept it because "i get what i pay for"...?

*zippin' thru the highway without a steering wheel*
(i do NOT drive a neon)

^^^ hahaha ... exactly.

do you think people never complain at mcdonalds? of course they do.

Originally posted by netdude
exactly... i mean... my only response is... if i buy a brand new dodge neon and the steering wheel comes off... what do i do? just accept it because "i get what i pay for"...?

*zippin' thru the highway without a steering wheel*
(i do NOT drive a neon)

This is complete BullCrap and you know. Without a steering wheel you cannot drive AT ALL. With Cogent going over 300ms and having slow downs, you can compare it to a car temporary not capable going over 60 cause the gasoline is not coming fluently but the "car" DOES still drive and has a steering wheel.

Cogent doesn't need you as a client so why don't you stuff it and pay Verio 90$/mbit and leve us Cogent folks alone.

ok... fine... i buy a dodge neon and it won't go over 30 without going putt-putt-putt then stalling... i just accept that? "i got what i paid for i guess... yup, all 4 tires are there... missing the spare, but i guess i got what i paid for"

Have you checked the wiring close to the EGR valve?

it can burn out and cause problems like that.


Originally posted by netdude
ok... fine... i buy a dodge neon and it won't go over 30 without going putt-putt-putt then stalling... i just accept that? "i got what i paid for i guess... yup, all 4 tires are there... missing the spare, but i guess i got what i paid for"

Originally posted by netdude
ok... i've got a server in chicago... (and yes, i use cogent only because they're cheap but if there were anybody as easily accessible, i'd drop cogent n run like hell from that cr*p)...



Please explain what you mean by 'as easily accessible'.

Am I the only one that feels that this thread is/has become very low? I have nothing to add to this nonsense. Ask me again if there is a REAL discussion going on and not a bashing... cu

Originally posted by NocSol
Am I the only one that feels that this thread is/has become very low? I have nothing to add to this nonsense. Ask me again if there is a REAL discussion going on and not a bashing... cu

If you had nothing to add to this nonsense what's the purpose of claiming that this thread is now low and not a "REAL" discussion? Isn't this "last" post of yours just adding to the fuel?

Excellent, one post closer to calling myself the Holy Father of WHT.

Originally posted by mushrew
Excellent, one post closer to calling myself the Holy Father of WHT.

(Don't forget the trademark...) *very important*

On with the thread...

Originally posted by rastoma


Please explain what you mean by 'as easily accessible'.

my definition of easily as accessible is generally that... there are other bandwidth providers out there that are reasonably priced (verio, level3 and yipes! primary)... if i could just find a dedicated server provider that'd offer a dedicated server with them including the high bandwidth allowance (in accordance with their rates), i'd leave cogent... but i have my own plans that i am pursuing... and once i do have my financing, i am definately leaving cogent for good...

i'm not very mechanical... i tried giving it a boot, but it only left a dent... next, i'm gonna swing a crow bar around the engine and see whatever it hits... i was thinking of taking it back to the dealership, but have decided that "i got what i paid for"...

(just to make it clear, i don't really drive a neon, lol)...

Originally posted by frnht451
Have you checked the wiring close to the EGR valve?

it can burn out and cause problems like that.

Originally posted by netdude


my definition of easily as accessible is generally that... there are other bandwidth providers out there that are reasonably priced (verio, level3 and yipes! primary)... if i could just find a dedicated server provider that'd offer a dedicated server with them including the high bandwidth allowance (in accordance with their rates), i'd leave cogent... but i have my own plans that i am pursuing... and once i do have my financing, i am definately leaving cogent for good...

well.. there are hundreds of hosts out there that started with and are still making HUGE amounts of money with connections that are NOT cogent only, so I still don't know what you mean by wanting something else easily accessible.

if you're talking about you can't offer 10 gigs of hard drive space and 100 gigs of bandwidth for $10/month from a QUALITY provider then the connection won't matter as you won't be in business 6 months from now.

if you want to be profitable and stay in business then get a quality provider and make your own mark on the hosting business.... and you don't make your own mark by copying everyone else but just give more bandwidth. people don't want more bandwidth (not the professionals in the world).. they want quality.