Slashdot Mirror

Source: http://www.dslprime.com/dslprime/42-d/4830-internet-transit-costs-down-50-in-last-year

i'd gladly pay $3.50 per megabit/sec for symmetrical, spook-free, unfettered and unmetered access... hell, give the isp a profit margin and make it $5.00 per megabit... it's WHAT WE PAY NOW for 15 meg, but that's just the download speed, and that's just for a measly 100 gigs a month (or, roughly the equivalent of 300 kilobits/sec 24/7.. far short of the 15 meg we're paying for)

Even if you take them at their word, bandwidth is not the highest cost component of an ISP's business. It is all in the infrastructure and that is basically fixed whether you use one 1 byte or 10 terabytes.

Over the last few years, wholesale IP transit costs have dropped 50% per year. Nowadays big ISPs are probably paying roughly $6 per terabyte. With pricing so cheap it is obvious that usage is not the driving cost.

Source: http://www.dslprime.com/dslprime/42-d/4830-internet-transit-costs-down-50-in-last-year
(I realize that ip transit is priced by data rate not total bytes, but all of these usage-based billing schemes are priced in bytes per month, so I did a rough conversion of the units in the source to the units comcast would use for pricing.)

Do you have reliable sources for that number?

http://www.dslprime.com/dslprime/42-d/4830-internet-transit-costs-down-50-in-last-year

I did the math to convert units from the way ip transit is priced to the way most isp users think about it.

You're making excuses: http://dslprime.com/a-wireless-cloud/61-w/4466-us-wireless-75-fewer-basestations-than-comparable-europe

You have no comprehension of how badly US telcos underinvest in their networks. In 2007 AT&T was building less towers than T-Mobile while ignoring their engineers' warnings about the network meltdown their actions would cause.

Lol @ infrastructure buildup. You are an incredibly naive individual. American has 75% fewer basestations that Europe: http://dslprime.com/a-wireless-cloud/61-w/4466-us-wireless-75-fewer-basestations-than-comparable-europe

Uh... they work out the costs: http://www.dslprime.com/dslprime/42-d/4148-costs-and-caps

Demand expands at a constant and predictable rate. In fact wireless annual rate of growth has decreased over the years. The real issue for AT&T was running fiber to their towers, instead of the pathetic hodgepodge of T1s that they've been using for years. AT&T has been experiencing record profits for years. They are completely able to handle any increase in traffic by upgrading their networks.

This is a good article to grasp some of what's going on: http://dslprime.com/a-wireless-cloud/61-w/2823-mobile-data-growth-down-to-60-from-5000

AT&T's congestion problems are self-inflicted, through a reduction in capex and network investment. http://www.circleid.com/posts/20100215_absolutely_no_wireless_spectrum_shortage_in_2010/

There is no wireless spectrum shortage, as admitted by Verizon's own CEO: http://dslprime.com/a-wireless-cloud/61-w/2844-no-spectrum-for-competition-why-verizon-turned-around

AT&T's congestion problems are self-inflicted, through a reduction in capex and network investment. http://www.circleid.com/posts/20100215_absolutely_no_wireless_spectrum_shortage_in_2010/

There is no wireless spectrum shortage, as admitted by Verizon's own CEO: http://dslprime.com/a-wireless-cloud/61-w/2844-no-spectrum-for-competition-why-verizon-turned-around

AT&T's congestion problems are self-inflicted, through a reduction in capex and network investment. http://www.circleid.com/posts/20100215_absolutely_no_wireless_spectrum_shortage_in_2010/

There is no wireless spectrum shortage, as admitted by Verizon's own CEO: http://dslprime.com/a-wireless-cloud/61-w/2844-no-spectrum-for-competition-why-verizon-turned-around

You have no friggin' idea what you're talking about. 12.7 million *homes* have FIOS. How many households exist in the the US?

And rural fiber is similar in cost to urban and suburban fiber. http://www.dslprime.com/fiber-news/175-d/1755-cost-of-rural-fibe-tv-1500-dropping

In fact the cheapest place to deploy fiber is in suburban areas, since urban areas are often so densely packed that running fiber into a building becomes a complex task of navigating narrow spaces in existing structures.

$24.95/month is "cheap"? It only gets you a "scabby" connection plan? 100 gigs/month is a lot?
You guys really need to get some new ISPs. Go check if Finland, South Korea or Hong Kong feels like exporting some to you!

Mandatory Filtering of Indecent and Obscene Material. M2Z commits to mandatory filtering of indecent and obscene material for the National Broadband Radio Service. This will be accomplished through a compulsory setting on the service that will utilize state of the art filters, taking every reasonable and available step to block access to sites purveying pornographic, obscene or indecent material. Like the free service itself, M2Z's content filtering will be "always on." Moreover, National Broadband Radio Service customers will be unable to alter the filters as they constitute an essential element of that service. To accomplish these critical filtering functions, M2Z plans to route National Broadband Radio Service traffic through a set of servers that can examine the traffic flows for improper activity and restrict access as required. Thus, the nation's children -- and their parents -- will have free access to broadband that is not only very affordable but also family-friendly and free from pornographic and other indecent material.

Boy, nobody seems to have read this very carefully -- including the ZDNET reporter wrote the original article.

Here are the relevant TI pages: their press release, and their UDSL home page.

First: the 200 Mbps peak bandwidth referred to is *aggregate* bandwidth -- upstream plus downstream -- for a symmetric 100/100 connection. These speeds have been talked about in the industry for some time -- see back issues of the DSL prime newsletter. The peak bandwidth in any one direction is probably 100 Mbps.

Second: The peak speeds are indeed at short distances, but they won't be limited to just those people living a block from their CO. Instead they will be delivered as FTTC (fiber-to-the-curb) or FTTP (fiber-to-the-premises), where an optical fiber runs from the CO to a local mini-DSLAM that serves a small neighborhood -- or even a single apartment building.

Third: TI is *not* selling UDSL as an ultra-high-speed DSL, but as a "universal" DSL to make it easier (read: cheaper) to upgrade customers when the time comes. Other silicon manufacturers will be selling similar technologies... But for the DSL customer, what matters is the "when the time comes" part -- and unless your neighborhood already has FTTC/FTTP, the time isn't going to come for a little while yet. :-)

But it will come.

All this stuff is down at the physical layer. There's no mention of the higher layers; apparently we're still stuck with PPPoE, a login mechanism, and client software.

The big win with this thing will be the improved diagnostics, along with slightly better noise immunity and the power-save modes.

The increases in performance and range are pretty minimal. An additional 50kbps and 600ft of range isn't all that impressive, although the fact that it is backwards compatible with some existing hardware is semi-promising.

Anyway, here's some extra info on ADSL2, or G.bis that i dug up:

http://www.aware.com/products/DSL/gbisadsl2.htm
http://www.convergedigest.com/Silicon/siliconartic le.asp?ID=5435
http://www.dslprime.com/a/adsl21.pdf(sorry about the pdf)

Qwest's Nacchio: VDSL is ready to make money
"VDSL is the killer for anybody with our kind of network"
Nacchio told analysts, "The technology works, and it is available at the right price." Anton Wahlman reports Qwest has prepared detailed plans to roll out VDSL in as many as 10 cities in a short time frame. He sees the cost as low as under $1,000 per subscriber depending on what is included in the calculation. They would soon pass 3M homes, a network similar in size to Cablevision. Installing 50,000 lines in Phoenix taught them how to do the job inexpensively (technician training is crucial), but Nacchio told us last year he had decided to wait until the next generation of equipment brought the costs down.

The presumed reason Qwest is now holding off is their dramatic drop in capex of $2B, accompanied by layoffs - not the time to announce new initiatives. Nacchio was more politic, instead announcing the only remaining problem was regulatory. Phoenix is treated by regulators as a "Title 6" (unregulated cable) and Nacchio wants a clear ruling that expanded VDSL would not be instead considered "Title 2" (telephony, subject to competitive and unbundling rules.) His clear implication, Wahlman reports, is that Qwest will go ahead if they get clearance. Motorola's Galvin believes him, and invested $20M more in Next Level. I think it's a good bet, although DSL Prime urges investors to be very careful buying NLC shares. Only a minority of shares remain in public hands, and it's always risky for minority holders.

For those who are interested in the capital (equipment) cost of rolling out a DSL service, the current basic cost is down to about $50 for a USB modem, and $100 for the DSLAM port. (assuming you want 10,000 of them).
Of course, you have to add in the warm bodies in the call center, and the backbone bandwidth, but these are telcos, and are supposed(!) to have a good ATM backbone already. (Besides, moving surfers from dial-up to DSL saves a *HUGE* volume of voice capacity on the backbone anyhow).
Source for the above figures is the latest news report from DSL Prime.
They provide a good free (as in beer) report on the DSL industry.

For those who are interested in the capital (equipment) cost of rolling out a DSL service, the current basic cost is down to about $50 for a USB modem, and $100 for the DSLAM port. (assuming you want 10,000 of them).
Of course, you have to add in the warm bodies in the call center, and the backbone bandwidth, but these are telcos, and are supposed(!) to have a good ATM backbone already. (Besides, moving surfers from dial-up to DSL saves a *HUGE* volume of voice capacity on the backbone anyhow).
Source for the above figures is the latest news report from DSL Prime.
They provide a good free (as in beer) report on the DSL industry.