dslreports logo
 story category
Cogent 'McBandwidth' Gets Cheaper
Company tries to distance itself from Level 3, Global Crossing
Cogent, already known as the "Walmart of bandwidth" (and no stranger to peering disputes) says they're taking their prices even lower in order to fend off competitors Level 3 Communications and Global Crossing. According to Telephony Online, Cogent this morning announced that customers who commit to three-year contracts and higher volume service provider customers will now get access to bandwidth at a flat $7 a megabit. Prices go as low as $4, for three-year contract customers who consume a 10 gigabit port. An interesting comment from the report:
quote:
Click for full size
"We have seen Internet traffic growth slow over the past year as measured by a couple of references," Schaeffer said. "The rate of growth in percentage terms has slowed and that is because of a number of factors. The casual video and social networking sites that drive a lot of traffic are maturing and we have not seen the huge wave of displacement by professional video services that would cannibalize cable and satellite TV."
view:
topics flat nest 

Jerm
join:2000-04-10
Richland, WA
·Ziply Fiber

1 edit

1 recommendation

Jerm

Member

Very Interesting ...

So much for cable and their whining/caps.

Thats about $0.012 (or 1.2 cents) per GB of transfer when I do the math (@$4/mbps fully utilized).

Just proves the caps and are a step backwards - the old mantra remains:

It is easier to add capacity to deal with a bandwidth problem than it is to mess with the alternatives

Can you imagine in 1998 if ISPs had said: "Well instead of providing faster pipes, we're gonna provide metered service because thats what our customers really want!"

Robert
Premium Member
join:2001-08-25
Miami, FL

Robert

Premium Member

Re: well...

said by Jerm:

hmm so much for Cable and their caps...
What does this have to do with cable providers and their caps?
88615298 (banned)
join:2004-07-28
West Tenness

88615298 (banned)

Member

Re: well...

said by Robert:

said by Jerm:

hmm so much for Cable and their caps...
What does this have to do with cable providers and their caps?

Jerm
join:2000-04-10
Richland, WA
·Ziply Fiber

Jerm

Member

Re: well...

said by 88615298:

said by Robert:

said by Jerm:

hmm so much for Cable and their caps...
What does this have to do with cable providers and their caps?

... at least someone gets it ...

It's like Verizon charging $0.15 per text msg. Or TWC and their proposed $1.50/GB fee.
patcat88
join:2002-04-05
Jamaica, NY

patcat88

Member

Re: well...

said by Jerm:

... at least someone gets it ...

It's like Verizon charging $0.15 per text msg. Or TWC and their proposed $1.50/GB fee.
Remember a cellular phone call is 7 to 11 text messages PER SECOND bandwidth wise. Nothing like getting payed half a million $ to move 1 CD worth of text messages. $930K per GB of texts. They must be laughing all the way to the bank.

Jodokast96
Stupid people piss me off.
Premium Member
join:2005-11-23
NJ

Jodokast96 to Robert

Premium Member

to Robert
said by Robert:

said by Jerm:

hmm so much for Cable and their caps...
What does this have to do with cable providers and their caps?
From the arguement that was made by some that a customer using their full bandwidth costs a cableco money, hence the need for caps.

Robert
Premium Member
join:2001-08-25
Miami, FL

1 recommendation

Robert

Premium Member

Re: well...

said by Jodokast96:

said by Robert:

said by Jerm:

hmm so much for Cable and their caps...
What does this have to do with cable providers and their caps?
From the arguement that was made by some that a customer using their full bandwidth costs a cableco money, hence the need for caps.
But that really isn't why caps exist. Caps exist because of the limited amount of bandwidth that a cable co can send to its customers. It's not about costing more money, it's about trying to give everyone an equal share.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

3 edits

Dogfather

Premium Member

Re: well...

Cable providers have no trouble providing an equal share at the backbone connectivity level just like telcos don't.

They do have trouble at the nodal level on occasion which requires traffic management or bandwidth saving measures that allow more channels to be assigned for transport between those saturated nodes and the head end.

It's this particular capacity problem that DOCSIS 3, SDV and other tricks will solve.

Robert
Premium Member
join:2001-08-25
Miami, FL

Robert

Premium Member

Re: well...

said by Dogfather:

Cable providers have no trouble providing an equal share at the backbone connectivity level. They do have trouble at the nodal level on occasion which requires traffic management or bandwidth saving measures that allow more channels to be assigned for transport between those saturated nodes and the head end.
Yea, that's what I meant.

djrobx
Premium Member
join:2000-05-31
Reno, NV

2 edits

djrobx to Dogfather

Premium Member

to Dogfather
quote:
Cable providers have no trouble providing an equal share at the backbone connectivity level just like telcos don't.

They do have trouble at the nodal level
I don't buy it. I didn't buy it back when Pacific Bell was pushing the "webhog" ads, either. Back then this same argument was being made, when cable's caps were as low as 1500/128.

Cable companies are now offering 10x those speeds on pretty much the same DOCSIS 1 technology, and overall it's holding up quite well.

Think about how many nodes cover a city, and then think about how much backbone capacity it would require to fill all of those nodes, at 38mbps per node. And don't forget, you need a very big link from the cable plant TO the backbone provider that can handle all of that traffic.

Of course there will be some clogs at the node level in certian active neighborhoods. I'm not saying it doesn't happen. There's always going to be choke points in the network that require maintenance.

Time Warner's SoCal region seems to have more capacity problems at the city level than they do the node level, as made evident by the threads that crop up every few months.

-- Rob

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

1 edit

Dogfather

Premium Member

Re: well...

Cable operators don't have problems if they dedicate enough channels to their HSI offerings.

It doesn't matter how many nodes cover a city, it's how many customers they're trying to serve with a node and how many channels they're dedicating to get that done.

It's no secret that there are markets with saturated channels otherwise you wouldn't see evening slowdowns in some parts of a city while the other parts run fine. And not every market has problems which is why I said they have problems on occasion.

I was a TWC 15/2 customer up until a couple of weeks ago and never had an issue, but I know people served by the same head end that did during the evening, particularly in upstream speeds and and occasionally latency.

And while cable systems for the most part are handling current traffic demands, cable operators can't continue ramping up speeds without providing an increase in node to headend capacity.

SpaethCo
Digital Plumber
MVM
join:2001-04-21
Minneapolis, MN

SpaethCo to djrobx

MVM

to djrobx
said by djrobx:

Back then this same argument was being made, when cable's caps were as low as 1500/128.

Cable companies are now offering 10x those speeds on pretty much the same DOCSIS 1 technology, and overall it's holding up quite well.
Back then it was also very common to share the same downstream channel across a half dozen nodes or more. Over the last decade the MSOs have been pretty active in getting the number of nodes per downstream channel reduced, down to 1 DS in many cases. They've also added multiple upstream channels and performed physical node splits to get the number of homes per node down.
said by djrobx:

Think about how many nodes cover a city, and then think about how much backbone capacity it would require to fill all of those nodes, at 38mbps per node. And don't forget, you need a very big link from the cable plant TO the backbone provider that can handle all of that traffic.
That infrastructure is in place for many carriers.

Look at Comcast's network:

The CMTS hardware is connected to user/utility routers with GigE connections. The user routers connect to the aggregation routers with multiple 10GigE connections, and the area/aggregation routers connect to the core with multiple 10GigE connections.

Comcast's network is extremely fat in the middle, with Internet access circuits provisioned by demand (at significantly less capacity than what is available in the core) and the CMTS routers can't generate enough traffic to come even remotely close to choking the GigE uplink connections they have into the network.

For MSOs the congestion is in the last mile that connects the subscribers, followed by a distant chokepoint at the Internet access points.

boils down
@mcleodusa.net

boils down

Anon

Re: well...

Great so what you're saying is that is has nothing to do with the wholesale cost of traffic to their upstream providers and more to do with them not being able to offer the speeds and services they advertise to more and more customers without charging overages under the guise of collecting more money for "upgrades" (As if somehow what users currently pay does not allow a cable provider to do anything more than scrape by). If it truly is not related to the extremely cheap cost of wholesale bandwidth, then i suppose they will just implement a "free nights and weekends" sort of deal where they are unmetered, or unrestricted during periods of non peak usage right?

OR

Being somewhat knowledgeable in the way corporate america operates what is the more logical conclusion?

en102
Canadian, eh?
join:2001-01-26
Valencia, CA

en102 to Robert

Member

to Robert
said by Robert:

But that really isn't why caps exist. Caps exist because of the limited amount of bandwidth that a cable co can send to its customers.
That's only 'part' of the equation.
Why would Cable companies require a cap on a 512kbps connection then ? Its not entirely about having a limited amount of bandwidth, or they would not be selling 10-20Mbps plans with 95GB cap and a 512kbps plan with a 5GB cap. There's a profit margin they want to keep. By having a low bandwidth cap, they can push people off lower tiers. Many don't need +6Mbps, but they may want more than 25GB/month. Eg. I have a 3Mbps plan ... should I only be limited to 15GB?

Jodokast96
Stupid people piss me off.
Premium Member
join:2005-11-23
NJ

Jodokast96 to Robert

Premium Member

to Robert
True, but that is the reason some (not me) have put forth as a need for caps.

Jerm
join:2000-04-10
Richland, WA
·Ziply Fiber

Jerm to Robert

Member

to Robert
said by Robert:

Caps exist because of the limited amount of bandwidth that a cable co can send to its customers. It's not about costing more money...
"Limited amount of bandwidth"??? Then UPGRADE the dang backbone! Most cable co's own their own regional fiber networks, upgrade the friggin lasers!. And DOCSIS 3 will take care of the last mile.

On the contrary, it's all about the money. Oh and preventing online video from competing. After all what takes the most bandwidth? hmm...

Robert
Premium Member
join:2001-08-25
Miami, FL

Robert

Premium Member

Re: well...

said by Jerm:

said by Robert:

Caps exist because of the limited amount of bandwidth that a cable co can send to its customers. It's not about costing more money...
"Limited amount of bandwidth"??? Then UPGRADE the dang backbone! Most cable co's own their own regional fiber networks, upgrade the friggin lasers!. And DOCSIS 3 will take care of the last mile.

On the contrary, it's all about the money. Oh and preventing online video from competing. After all what takes the most bandwidth? hmm...
It's not about the backbone. It's about the last mile, and yes it's about money. But it's not about the Cogent lowering their pricing.

SpaethCo
Digital Plumber
MVM
join:2001-04-21
Minneapolis, MN

SpaethCo to Jerm

MVM

to Jerm
said by Jerm:

And DOCSIS 3 will take care of the last mile.
DOCSIS 3.0 will help the last mile, but only once they upgrade every installed CMTS and replace every single subscriber cable modem with a DOCSIS 3.0 model.
markopoleo
join:2003-04-02
Bonne Terre, MO

markopoleo

Member

Re: well...

said by SpaethCo:

said by Jerm:

And DOCSIS 3 will take care of the last mile.
DOCSIS 3.0 will help the last mile, but only once they upgrade every installed CMTS and replace every single subscriber cable modem with a DOCSIS 3.0 model.
Will you people stop thinking DOCSIS 3.0 is the answer to all.

It means NOTHING to the end user currently, it just helps the cable provider.

Let me say it again..current cable customers have NOT REACHED THE BANDWIDTH OF CURRENT DOCSIS MODEMS. Had to yell it cause some people are not listening.
Lazlow
join:2006-08-07
Saint Louis, MO

Lazlow

Member

Re: well...

Markopoleo

You forget about the shared part of the Docsis. We all (on the same node) have to share the bandwidth on the channel(38 down and 10 up for Docsis 1.1, most markets). You start splitting up 10 (up) with a bunch of guys running 16/2 and the brakes go on in a hurry.
markopoleo
join:2003-04-02
Bonne Terre, MO

markopoleo

Member

Re: well...

said by Lazlow:

Markopoleo

You forget about the shared part of the Docsis. We all (on the same node) have to share the bandwidth on the channel(38 down and 10 up for Docsis 1.1, most markets). You start splitting up 10 (up) with a bunch of guys running 16/2 and the brakes go on in a hurry.
That really is not a DOSIS problem though, thats just a problem in deployment. They should not even be oversubscribing nodes in the first place. Nothing saying they won't use 3.0 and oversubscribe it anyways even if they have more bandwidth at headend.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

Dogfather to markopoleo

Premium Member

to markopoleo
said by markopoleo:

said by SpaethCo:

said by Jerm:

And DOCSIS 3 will take care of the last mile.
DOCSIS 3.0 will help the last mile, but only once they upgrade every installed CMTS and replace every single subscriber cable modem with a DOCSIS 3.0 model.
Will you people stop thinking DOCSIS 3.0 is the answer to all.

It means NOTHING to the end user currently, it just helps the cable provider.

Let me say it again..current cable customers have NOT REACHED THE BANDWIDTH OF CURRENT DOCSIS MODEMS. Had to yell it cause some people are not listening.
Huh? DOCSIS 3 absolutely is the answer to increasing speeds to end users. The biggest benefit is channel bonding.

And yes, there are cable systems where channel capacity is saturated. The bandwidth capacity of the modem is IRRELEVANT in cable topology. Those modems share bandwidth to the head end and without more dedicated channels, you aren't going to get any more speed, no matter what the modem's capacity is. And without channel bonding you will NEVER see the speeds to the end user that a competitor like Verizon is capable of delivering.

tiger72
SexaT duorP
Premium Member
join:2001-03-28
Saint Louis, MO

tiger72 to Robert

Premium Member

to Robert
said by Robert:

But that really isn't why caps exist. Caps exist because of the limited amount of bandwidth that a cable co can send to its customers. It's not about costing more money, it's about trying to give everyone an equal share.
That was their argument for Docsis... Anyone remember @home in the pre-docsis days? THAT's when people really shared and fought for bandwidth. Docsis allows everyone an equal share.

Caps are a "solution" to oversubscription.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

2 edits

Dogfather to Jerm

Premium Member

to Jerm

Re: Very Interesting ...

Cable's capacity problem isn't their backbone connectivity. It's between the node and the headend where they've dedicated too few channels to HSI, particularly in the upstream direction. That's why they're working on SDV, DOCSIS 3, ditching analog video and other bandwidth saving measures within the local cable systems.
EPS4
join:2008-02-13
Hingham, MA

EPS4

Member

Re: Very Interesting ...

There's also talk about backbone... though that seemed to be more from at&t than the cablecos.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

Dogfather

Premium Member

Re: Very Interesting ...

Backbone connectivity savings (eg P4P and other projects) is just about saving money, not because there is an actual capacity problem.

Obviously the less data an ISP can get away with buying the better it is for them. But AT&T wouldn't be capping because of capacity...it's 'cause they're greedy scumbags following the cable whores down the price increase toilet.

In the short term cable has capacity issues within individual cable systems so in some circumstances the excess traffic is causing them capacity problems, but quickly deploying technologies like I mention above takes care of that.

Any caps are simply about increasing revenue, not because caps are necessary to maintain network stability.

Jerm
join:2000-04-10
Richland, WA
·Ziply Fiber

Jerm

Member

Re: Very Interesting ...

said by Dogfather:

Any caps are simply about increasing revenue, not because caps are necessary to maintain network stability.
Bingo. And don't forget preventing comptition from online video - after all the cable co's are terrified of becomming a "dumb pipe" provider and having their cash cow drift away like VOIP is doing to the telcos...

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

1 edit

Dogfather

Premium Member

Re: Very Interesting ...

That's true. I'm watching a South Park episode via the Internet based DirecTV On Demand while I'm downloading Semi-Pro in 720P on my AppleTV. These services were once exclusive domain of cable companies.

JamesPC
join:2005-10-12
Orange, CA

JamesPC to Dogfather

Member

to Dogfather
Agreed Skeedatl.
jc10098
join:2002-04-10

jc10098 to Dogfather

Member

to Dogfather
However, aren't all ISPS hooked into the backbones and simply resetting the bandwidth that level 3 providers sell? My understanding, and I might very well be wrong, is that ISPS wire neighborhoods and business. However, the actual capacity is the backbone on which the ISP leases the bandwidth. So lets say X company has a backbone in LA. This backbone has 1 10 gigabits of speed for say 100,000 people. The slow downs come w here an ISP has too many people sharing a node, but not because of said backbone. The capacity is there, just the fact an ISP doesnt have enough nodes in the area. It'd be like a 6 lane highway being able to handle those 100,000 people until construction forces it down to 2 lanes a mile before the highway ends. The highway itself is sound up until the point of exit.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

4 edits

Dogfather

Premium Member

Re: Very Interesting ...

said by jc10098:

However, aren't all ISPS hooked into the backbones and simply resetting the bandwidth that level 3 providers sell?
Not always. The largest providers are tier 1s themselves like Verizon and AT&T. Time Warner also owns a massive network.

Backbone connectivity is scalable and bandwidth is cheap and evidentally getting cheaper.

•••

djrobx
Premium Member
join:2000-05-31
Reno, NV

djrobx to jc10098

Premium Member

to jc10098
quote:
The capacity is there, just the fact an ISP doesnt have enough nodes in the area
Don't forget about the intermediate connections between the cable plant and the backbone peering point. Such connections need to be able to handle the traffic from every node.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

Dogfather

Premium Member

Re: Very Interesting ...

It doesn't matter how many nodes a cable system has. Nodes themselves can handle a boatload of traffic.

The capacity crunch is solely how many channels a cable operator dedicates for traffic between that node and the headend. You can handle hundreds and hundreds of customers on a node, if you dedicate the channels to do it.

Backbone capacity is easily scaleable and because the pool of customers is so much larger, the backbone capacity isn't suceptible to overselling issues like nodal capacity is. It only takes a few seeding customers to saturate the upstream channels of a node.

HEDP
join:2008-04-27
Miami, FL

1 edit

HEDP to Jerm

Member

to Jerm
A T3 fully utilized at 45mbps up/down costs you around 4K-5K per month from Cogentco.

Ethernet can vary from 2K-4K with faster speeds, it also depends if you are on network or off network and how far you are from their POP, oh and that's per month.

A cable company asks for 150 dollars for a 18mbps line with I am sure but I think it was 45GB cap, I say from what Cogentco is charging you without a email account, without newsgroup access you are getting a bargain as a consumer for that pipe.

Also I am sure those prices are per data transferred, not the cost of speed.

If you don't believe me give Cogentco a call. I just did, to confirm my facts.

Dogfather
Premium Member
join:2007-12-26
Laguna Hills, CA

3 edits

Dogfather to Jerm

Premium Member

to Jerm
Well it certainly shows that these overage charges are price gouging to the Nth degree.

Like Time Warner, certainly with their volume they get great prices while charging the customer $1.50/GB.

SpaethCo
Digital Plumber
MVM
join:2001-04-21
Minneapolis, MN

SpaethCo to Jerm

MVM

to Jerm
said by Jerm:

So much for cable and their whining/caps.

Thats about $0.012 (or 1.2 cents) per GB of transfer when I do the math (@$4/mbps fully utilized).
Keep in mind that pricing is only valid in a Cogent meet-me facility -- you have to pay to get your own circuit to their handoff.

I've been going through the pricing exercise of getting OC192 connectivity between our data centers in Minneapolis, MN and Trumbull, CT. From the 511 building downtown (carrier neutral facility in Minneapolis) to a carrier neutral facility in Hartford, CT I can get 2 x OC192 circuits for $5k/mo each.

The carrier tail circuits, however, are about $20-35k/mo on the Minneapolis side (depending on private fiber or LEC circuit) and $25-45k/mo on the CT side for each circuit from Hartford to Trumbull.

They can sell cheap connectivity from major city to major city because the fiber is already in place, there's lots of it, and on top of it all they can "colorize" into DWDM bundles and get up to 64 connections on a single fiber pair pretty cheaply. It's getting the connections from these places of mass aggregation out to your location that drives up the costs.

••••••
raccettura0
join:2002-09-28
USA

raccettura0 to Jerm

Member

to Jerm
said by Jerm:

So much for cable and their whining/caps.

Just means more profit. Cable providers connect to upstream provider like Cogent. If they lower the cost, and cable companies cap their service (hence cap their costs)... that results in higher profit margins.

Just more reason to go ahead.
tmc8080
join:2004-04-24
Brooklyn, NY

tmc8080

Member

domestic vs international

Isn't cogent tier-1 for international traffic & tier-2 for domestic (US) traffic? This is more about getting data shipped to & from overseas.. but for some reason Cogent seems to think that has peaked & is on the decline. Companies engaging in multi-source hosting aren't going to be enticed to stream unless it's worth their while. On the other hand, demand will only be there if two things materialize.. demand for content increases & last mile providers raise broadband caps, instead of trying to bill by the byte.

This of course highlights some real problems in the international economy. As energy prices rise, consumers cut back & companies become more desperate to find ways to make a profit. Just don't run into any business model icebergs trying to make it work. Without companies such as Cogent, more providers would be chomping at the bit to bill by the byte.

SpaethCo
Digital Plumber
MVM
join:2001-04-21
Minneapolis, MN

SpaethCo

MVM

Discounted price for partial access

Interesting Cogent would take the discount route instead of improving their network, but they did lose a good chunk of customers after the recent Telia depeering. Given that they've been depeered a few times in the past, with Level(3) and Telia being the most recent I would be hesitant to sign a 3 year agreement with them even if they were used as only one of several upstream providers.

Latency performance with Cogent isn't always ideal either as they are missing large sections of mid-continental routing. Want to go from Chicago to Jacksonville? They don't have a link between Chicago and Atlanta like most carriers; instead you get to take the path all the way up through Boston and then back down the coast.

traceroute to www.colo4jax.com (38.113.65.18), 30 hops max, 40 byte packets
 1  69.65.40.62 (69.65.40.62)  0.356 ms  0.351 ms  0.810 ms
 2  so2-0-0-0.er1.Chi1.Servernap.net (69.39.239.169)  1.515 ms  2.012 ms  2.010 ms
 3  gi0-8.na21.b000359-1.ord01.atlas.cogentco.com (66.28.21.17)  2.006 ms  2.506 ms  2.503 ms
 4  gi0-7.3574.core01.ord01.atlas.cogentco.com (38.112.38.45)  2.484 ms  2.484 ms  2.482 ms
 5  po14-0.core01.alb02.atlas.cogentco.com (154.54.1.57)  19.780 ms  19.778 ms *
 6  po14-0.core01.bos01.atlas.cogentco.com (66.28.4.109)  22.722 ms  22.734 ms  22.722 ms
 7  po0-0.core02.jfk02.atlas.cogentco.com (154.54.7.78)  27.869 ms  28.081 ms  28.065 ms
 8  po13-0.core01.phl01.atlas.cogentco.com (66.28.4.2)  30.116 ms  30.259 ms  30.250 ms
 9  po5-0.core01.dca01.atlas.cogentco.com (66.28.4.17)  33.856 ms  33.857 ms  33.848 ms
10  po10-0.core01.atl01.atlas.cogentco.com (154.54.7.153)  46.148 ms  46.252 ms  46.241 ms
11  po6-0.core01.jax01.atlas.cogentco.com (154.54.3.198)  52.655 ms  52.684 ms  53.157 ms
12  vl3501.na21.b005947-0.jax01.atlas.cogentco.com (38.112.34.230)  52.651 ms  53.149 ms  53.144 ms
13  Tical_Network_Solutions.dmarc.cogentco.com (66.28.230.34)  52.643 ms  53.125 ms  53.126 ms
14  shellx.tical.net (38.113.65.18)  52.602 ms  52.605 ms  52.602 ms
 

••••••••

SLD
Premium Member
join:2002-04-17
San Francisco, CA

1 edit

SLD

Premium Member

Pricing is actually lower

We just re-signed for $6/Mbit on a 3-year for a Gig-E, so it is a dollar lower than the story says. The $7 is for bursting (used to be $15).

Cogent is pretty good - we are single-homed with them and our sites push 250Mb/sec avg. The only issues we've seen are a couple customers in the Nordic region (i.e. Telia) and now Comcast with their screwed up routing via GBLX which is overloaded (not Cogent). This should be resolved shortly.
nriley
join:2008-06-12
Champaign, IL

nriley

Member

Re: Pricing is actually lower

Where'd you see that Comcast's routing is going to be cleared up? I'm trying to get about 2 miles in the real world. With Insight my traffic went to Chicago; with Comcast/Cogent it used to just go to the east coast, but now it's just all over the place.

 3  ge-1-1-ur01.champaign.il.chicago.comcast.net (68.85.178.57)  8.989 ms  8.118 ms  9.124 ms
 4  pos-0-10-0-0-ar01.indianapolis.in.indiana.comcast.net (68.86.90.174)  18.508 ms  17.903 ms  22.118 ms
 5  pos-0-0-0-0-cr01.nashville.tn.ibone.comcast.net (68.86.90.173)  28.889 ms  31.591 ms  28.584 ms
 6  pos-0-12-0-0-cr01.newyork.ny.ibone.comcast.net (68.86.85.21)  42.642 ms  51.851 ms  40.201 ms
 7  te-0-4-0-4-cr01.losangeles.ca.ibone.comcast.net (68.86.84.129)  74.862 ms  76.133 ms  77.317 ms
 8  te-1-1-pr01.600wseventh.ca.ibone.comcast.net (68.86.84.130)  77.835 ms  81.022 ms  76.945 ms
 9  68.86.89.158 (68.86.89.158)  76.443 ms  75.527 ms  74.208 ms
10  vl3492.mpd01.lax01.atlas.cogentco.com (154.54.3.9)  74.624 ms vl3493.mpd01.lax01.atlas.cogentco.com (154.54.6.229)  76.855 ms vl3492.mpd01.lax01.atlas.cogentco.com (154.54.3.9)  77.390 ms
11  te4-2.ccr01.sna02.atlas.cogentco.com (154.54.3.37)  77.180 ms te2-4.mpd01.iah01.atlas.cogentco.com (154.54.5.101)  78.868 ms te8-4.mpd01.iah01.atlas.cogentco.com (154.54.2.197)  79.083 ms
12  te2-1.mpd01.dfw01.atlas.cogentco.com (154.54.5.133)  87.458 ms te7-3.mpd01.dfw01.atlas.cogentco.com (154.54.5.129)  84.593 ms te9-3.mpd01.dfw01.atlas.cogentco.com (154.54.25.97)  85.687 ms
13  te4-2.ccr01.phx02.atlas.cogentco.com (154.54.7.85)  87.612 ms te2-3.mpd01.mci01.atlas.cogentco.com (154.54.5.14)  123.445 ms te4-2.ccr01.phx02.atlas.cogentco.com (154.54.7.85)  86.350 ms
14  te4-2.ccr01.aus01.atlas.cogentco.com (154.54.1.165)  105.731 ms te9-3.mpd01.ord01.atlas.cogentco.com (154.54.25.82)  118.727 ms  117.586 ms
15  te8-2.mpd01.iah01.atlas.cogentco.com (154.54.1.66)  78.755 ms vl3605.na31.b002329-4.ord01.atlas.cogentco.com (38.20.36.178)  117.579 ms vl3905.na31.b002329-4.ord01.atlas.cogentco.com (38.20.36.186)  123.905 ms
16  te9-3.mpd01.dfw01.atlas.cogentco.com (154.54.25.97)  84.528 ms university-of-illinios-urbana.demarc.cogentco.com (38.104.99.42)  79.710 ms  81.593 ms
17  t-ur2rtr.ix.ui-iccn.org (72.36.126.101)  83.155 ms te2-3.mpd01.mci01.atlas.cogentco.com (154.54.5.14)  123.130 ms t-ur2rtr.ix.ui-iccn.org (72.36.126.101)  91.782 ms
18  iccn-ur2rtr-uiuc2.gw.uiuc.edu (72.36.127.6)  81.912 ms te9-3.mpd01.ord01.atlas.cogentco.com (154.54.25.82)  118.862 ms iccn-ur2rtr-uiuc2.gw.uiuc.edu (72.36.127.6)  82.789 ms
19  vl3605.na31.b002329-4.ord01.atlas.cogentco.com (38.20.36.178)  119.366 ms t-exiteb.gw.uiuc.edu (130.126.0.73)  83.016 ms  83.159 ms
20  * * *
21  t-ur2rtr.ix.ui-iccn.org (72.36.126.101)  84.330 ms *  82.494 ms
22  * * *
23  t-exiteb.gw.uiuc.edu (130.126.0.73)  81.796 ms  83.553 ms *
 

The reverse path is quite a bit shorter, but looks like it's somewhat overloaded as you mention.

 7  t-dmzo.gw.uiuc.edu (130.126.0.70)  0.708 ms  1.052 ms  1.006 ms
 8  ur1rtr-uiuc.ex.ui-iccn.org (72.36.127.1)  0.658 ms  0.977 ms  0.563 ms
 9  t-710rtr.ix.ui-iccn.org (72.36.126.110)  3.573 ms  3.528 ms  3.224 ms
10  i2-cps.ex.ui-iccn.org (72.36.127.138)  8.348 ms  3.485 ms  3.849 ms
11  ge3-32-1000M.ar4.LON3.gblx.net (64.208.110.189)  4.029 ms  4.075 ms  3.787 ms
12  te2-1-10G.ar5.CHI1.gblx.net (67.17.106.133)  4.706 ms  4.318 ms  3.945 ms
13  COMCAST-IP-SERVICES-LLC-Chicago.TenGigabitEthernet2-3.ar5.CHI1.g (64.209.88.254)  42.826 ms  45.647 ms  42.922 ms
14  te-3-2-ar01.area4.il.chicago.comcast.net (68.87.230.9)  43.326 ms  43.619 ms  43.016 ms
15  68.85.176.2 (68.85.176.2)  68.044 ms  67.542 ms  67.249 ms
16  c-5-0-ubr01.gardencity.mi.michigan.comcast.net (68.85.176.1)  74.565 ms  74.381 ms  74.433 ms
 

SLD
Premium Member
join:2002-04-17
San Francisco, CA

2 edits

SLD

Premium Member

Re: Pricing is actually lower

I'm keeping in contact with the Cogent NOC (and my sales rep) about the issue. That is quite an outbound route through TN, NY, LA via Comcast. Even though you have quite a few hops through Cogent after, it looks to be a fairly straight path through Houston and Dallas... but interestingly enough, I didn't see a single Cogent hop for the return path. I wouldn't be surprised if most of your latency are actually in the return path with GBLX though ...

NetAdmin1
CCNA
join:2008-05-22

NetAdmin1

Member

The down side...

The only problem with budget carriers like Cogent dropping their prices is that it attracts more and more fly-by-night hosting operations which host more than the world's share of rooted boxen, spam relays and warez/kiddie porn/etc dump sites....
GPSrob
join:2007-05-21

1 edit

GPSrob

Member

cost of internet bandwidth != cost of service delivery

Surely they're only offering those rates for the internet connected bandwidth... I think you still have to get a private line transport to one of their POPs or do colo arrangements. I guarantee you that your LEC, CableCo, LD company (long haul data), etc will not sell you a point-to-point DS3 or faster at anywhere near that price. Maybe by a multiple of 15-20 just for the transport... Still interesting though, because it makes connecting to other ISPs vastly less appealing.

JF Mezie
@teksavvy.com

JF Mezie

Anon

Deadpool they're calling your name

Deadpool even you can't deny facts and no one in the entire world is going to buy your spiel about bandwidth costing a fortune. Read it and weep and wake up to the real world in 2008.