tell me, what's wrong with cogent
 

i need to make some decisions, and i want to know what exactly, is it about cogent that makes it inferior.. everyone knows they're substandard, but i want it quantized.

i want to know things like: inferior infrastructure? higher latency? more hops? less redudancy? more 'hiccups'? or...??? (references to stats/articles would be appreciated)

has anyone (individual, magazine, review site, etc) performed an indepth analysis and comparision of cogent vs Tier 1?

another thing.. cogent typically comes with a 99.5% SLA... does anyone have any information as to the frequency and/or duration of downtimes?

You cheap bastard ;)

J/K! If it's just for gallery hosting I don't see anything wrong with it.

DH

i don't see u for months, and the first words outa your mouth are 'u cheap bastard'... lol

Frankly, I'd have no issue running non mission critical components. Things like hosted galleries, large movie files, that sort of thing that's going to take a beating and not return enough revenue to justify expensive bandwidth. Would I run my affiliate program or database applications on it? No.

Cogent
 

Well hope you got some time. I was one of the first ISP's to buy cogent, a gig of it, then 2. If you are strctly cogent to cogent their network rocks. The fastest cross country hop I ever saw was from cogent frisco to cogent dc. 10 ms, most isps are 60 milliseconds. So whats the issue? The issue is that they angered every tier 1 provider and so they peer with the tier 1's at public peering points. Generally these are over saturated. At times, because cogent has very few pull customers (those where the bulk of their bandwith is pulled by web surfers) and many push (web hosting companies) Other ISP's will make them pay for peering instead of the free peering tier 1's give each other. This peering is done privately. At different times the peers cogent has are so over subscribed that they are trying to fit 4 gigs of traffic over a .622 gig pipe. Other times the way they deploy their loops to people gets fucked. The cogent 1Gbps loop is made of of 2 oc-12 pipes (622 mbps) bonded on a switch . Lots of times they get short on equipment and all you get is 622 instead of 1000. There are other factors, inferior equipment, specifically the cisco 2948 l3G is a piece of shit and they use it a lot. Cisco after all funded them. The better switches at lower prices like foundry, extreme and riverstone are not available to them because of an exclusive agrreement. So hope that answers your question.

sorry, can't help you with that. m not familiar with cogent, whatever...

http://scoreboard.keynote.com/scoreb...assword=public

I remember a time when all the cogent times were in the "warning" range. Not sure if this means anything though.

I've been running my paysite with almost 1000 members on cogent for a couple of years now, not a single problem and not a single complaint from any member :)

Yep. Plus the fact that it survived the post-dot bust era (when others were doing a mass die off) tells you something about the company and its business strategy.

cogent is doing good now so i presume you won't have a problem taking their service.

I had very bad luck with cogent.
but this is ebcause i was realy stupied (paysites on cogent)...never never never use it for paysites.

interesting replies... thx for the details guys..

cogent APPEARS to be right up there with most Tier 1 providers... even my little informal tracert tests from multiple sources to both cogent and non-cogent boxes yield latency times which are extremely compatible...

from this experiment, the only noticeable difference was hop number... mebbe 10-20% more.. but given my small sampling size, could be just noise...

i understand there are companies out there that conduct performance evaluations of this nature... ie, in most major cities, they'll subscribe to one, if not several, ISPs.. they then conduct their tests and can give an accurate assessment of network health... does anyone know, or heard of these companies?

look on traceroute.com i think, if you are lookin for cogent or tier 1 hosting i have very low rates for you. see my spam below.

New pricing for Hosting as low as a Nickel per Gigabyte
22 Gigs to the net
5 data centers in LA
2 Data Centers in Bay Area
2 Data centers in Seattle/Vancouver
100% uptime if you choose multi-homed server packages

All single dedicated P4/AMD servers 99.00
200 GB BW included 99.99% SLA Guarantee
Free power rebooters supplied, manual reboots also free
windows or linux

Additional bandwith as follows:
Tier 2 (cogent and others in EQX peering exchange)
320 Gb (1 Mbps @ 50/M) = 50.00 only 16 cents per GB
3200 Gb (10 Mbps @ 40/M) = 400.00 only 13 cents per GB
8000 Gb (25 Mbps @ 30/M) = 750.00 only 10 cents per GB
16000 GB (50 Mbps @ 20/M) = 1,000.00 only 07 cents per GB
32000 GB (100 Mbps @ 15/M) = 1,500.00 only 05 cents per GB

Tier 1 (UUNET, Level3, Global Crossing, Teleglobe, AboveNet and more)
320 Gb (1 Mbps @ 75/M) = 75.00 only 23 cents per GB
3200 Gb (10 Mbps @ 50/M) = 500.00 only 16 cents per GB
8000 Gb (25 Mbps @ 35/M) = 875.00 only 11 cents per GB
16000 GB (50 Mbps @ 30/M) = 1,500.00 only 09 cents per GB
32000 GB (100 Mbps @ 25/M) = 2,500.00 only 08 cents per GB

All Data centers offer every bandwith solution!!!!!
Ask for custom routing and multiple subnet hosting!!!!
have Ip's in multiple Class C networks!!!!
Wholesale pricing for carriers!!!!