Slashdot Mirror
How Much Does a New Internet Cost?
wschalle writes "Given the recent flurry of articles concerning ISP over subscription, increasing bandwidth needs, and lack of infrastructure spending on the part of cable companies, I'm forced to wonder, what is the solution? How much would a properly upgraded internet backbone cost? How long would it take to make it happen? Will the cable companies step up before Verizon's FiOS becomes the face of broadband in America?"
It will always cost as much as you are willing to pay, and the upgrade does not matter here at all.
As long as they can get away with offering sub-par connectivity at premium prices, what incentive do they have to rock the boat? The only thing that can induce these telcos to make costly infrastructure upgrades is competition, which is in pretty short supply currently.
Yo dawg, I heard you like the Ackermann function, so OH GOD OH GOD OH GOD
If the money was not being spent on the war it will have been spent on something else, certainly not the internet backbone.
Linux Zealots: Smarter than Mac Zealots, but still zealots.
"What kind of economy do we want? And, what do we need to achieve it?"
Free market. End government supported monopolies to the extent possible.
I don't see why a private company doesn't set up a city-wide 802.11 wireless network. Businesses and private owners would be likely to let the company use the very small space required for the equipment, since customers would find wireless access attractive. Vending machines operate on this kind of principle, and there is no shortage of those.
It's nice to think that government could take care of the infrastructure instead, but do you trust the same people who can't fix potholes in asphalt with managing and maintaining a wireless LAN?
I don't, especially since after the network is installed, there's no political gain for maintaining it. It's the same reason great sysadmins whose systems never fail are typically seen as unnecessary.
If moderation could change anything, it would be illegal.
This Ask Slashdot question makes the false assumption that there is one, and only one Internet backbone, and that the only way to upgrade is to replace it. As Foldoc shows, the so-called backbone is composed of a number of large-scale networks that interconnect. If you need more bandwidth, all that's needed is to add as much as you need and can afford.
Good, inexpensive web hosting
So in other words, spending money on something stupid is ok because it would have been used on something stupid anyway? I realize trying to stop the government to spend money on useless think is an enormous game of whack-a-mole, but even I usually don't get that depressive.
Live today, because you never know what tomorrow brings
I had to quote that word because it's getting ridiculous how often it's thrown around now.
Anyway, the government should make, lay, and lease the fiber to the service providers, or even create one themselves. It would provide a MAJOR employment boost for the people, most notably the linemen who would actually lay the fiber. The manufacturing of it isn't rocket science and from the top down you could hire people for it, from the designers to the janitors. Teams of men and women would go out and work on the network and that would probably be thousands of jobs, if only temporarily. Keep some on per region (or many depending on how hard it is to upkeep) and keep the manufacturing plants open to sell the fiber to businesses.
Lay it all out like we did the highway systems, charge Verizon, Time Warner et. al. to use it. If it breaks, it's like a pothole, fix it.
Make it a not for profit (as if the government wasn't already) take all money from it and put it back into the network, not into some bridge to no where.
Upgrade as necessary, keep the country moving forward, the internet is too important to the world to allow it to slow or crash (not that I fear a crash).
My name is Anonymous Coward and I am running for President.
Yes, people are starving. Yes this site is US centric. Still, can't we have a tech discussion without the "people are starving" bullshit? Quit posting on slashdot and go feed them if its so important to you jackass.
you can't ack before you balls.. you just
The US DSL/cable/etc business model is built on a certain amount of oversubscription, just like (nearly) every network out there. I have worked for several companies, up to top 10 of the Fortune 500, and not a single one of them had a network that wasn't oversubscribed to a certain degree... even on the LAN (which is where it's the cheapest).
Those of you that work in a corporate environment with any density (>20 users on a floor, more than one floor)... If you've got a gigabit LAN, go ask your network guy if they have a 10-gig uplink for every 10 ports on the floor.
.
.
.
After he stops laughing and realizes you're serious, ask him why they are running an oversubscribed network. If he's on the design side, he'll end up telling you that you don't build a network for that level of traffic if it simply doesn't use it (most don't). The most likely place you're going to see a fully non-oversubscribed network is one that supports a supercomputer with many nodes. Even then you might see some.
It's just not economically feasible to build non-oversubscribed networks. Any of you know how much a card for a Cisco GSR that has just two OC-192 intermediate-reach ports on it is? MSRP is $585,000.
$585K for two 10 gigabit intermediate reach ports. And to build a non-oversubscribed network for a small community with say 2000 users on 8-meg cable connections that cost $60 a month. Gotta pay for the cable plant itself (to a certain degree), the fiber to link the customer-facing nodes (how much it cost to dig/hang/lay the fiber), the routers in the customer-facing nodes, the cards in the routers in those nodes (more bandwidth = higher cost cards), the distribution routers that link all the customer nodes together (and their cards), core routers with higher-speed interfaces to tie it all together if you have any decent number of distribution nodes (and their cards), peering routers to your upstream bandwidth provider (and cards), maintenance on every router/switch (which runs ~20-30% yearly over and above the purchase price), spares of a few of your most commonly-failing equipment, datacenter space, AC, cooling, engineering staff costs, field maintenance staff costs, systems administrators staff costs, 24x7 NOC staff costs, 24x7 helpdesk costs, multiple layers of management (each of those fields has to have management in an organization of any size), training costs to keep up on the latest developments, staff turnover costs, taxes... and that's before we've paid for one bit of peering bandwidth or even thought about making a profit - or considered what Mother Nature, backhoes, or out of control drunk drivers do to the equipment and fiber that make up the customer-facing network that sits in equipment sheds on concrete pads on the side of the road. And don't forget to add another 100% or so to all of those equipment costs, for redundancy. Don't want the whole east side of the city down because one port/device/fiber failed, do you?
There's a lot more than just a couple of Linksys gig switches and some cable RF converters that make up a cablemodem network. There's more than just a card in a phone switch that makes up a DSL network. The gear is very expensive, typically because there's lots of R&D that must go into the boxes to make them able to do what they do without having horrendous failure rates (which still happens sometimes).
France is an example of how different public policy decisions produce different outcomes.
France is pursuing, roughly, the public policy that the US adopted in the mid-1990s: Unbundle the local loop, permit competitive interconnection, encourage competition for services over the incumbent's old wire. That was, in fact, the gist of the Telecom Act of 1996.
In 2001, the Cheney-Rove regime's new FCC executed an about-face. They decided that the Bells were to be the winners, And their competitors were not to be. Furthermore, the Bells saw the Internet as the real enemy, not local telephone competitors per se, so they were allowed to execute their strategy to knock off the ISPs while replacing it with their own marginal substitutes. The last stage, which has not yet happened, is to remove "neutrality" from their networks, replacing Internet access with a set of "broadband services" of their own, like kickback-selected shopping, censored "news", and pay-per-view "media" access. That could never happen with real competition. The FCC's excuse is that there's cable, and a duopoly is "enough" competition, especially with the imaginary "third pipe" that never really appears in any useful way.
France, in contrast, stayed the course. There are multiple ISPs sharing the old FT wire. So advances in DSL technology meant advances in available speeds, and reductions in DSLAM prices and backbone ISP rates meant reductions in DSL charges. It's not exactly peaches and cream for FT, but it's great for the economy as a whole.
For something as important today as the Internet, it's surprisingly fragile and primitive. It's amazing we've gotten this far; it's not clear that "more of the same" can happen.
;-)
One obvious problem, at least in the United States, is the "last mile" or if you prefer "first mile" problem. In maybe half of homes it's a cable/ILEC (old monopoly phone company) duopoly. Most of the rest can get cable or telco DSL. A fair share can't get either yet. FCC statistics are intentionally deceptive about this, counting ZIP codes that have even one "broadband" subscriber as being served, even if most of the area isn't. And their 200 kbps downstream definition of "broadband" is pathetic.
DSL is a mid-life kicker for old copper. Passive Optical Network-style fiber, as in FiOS, is also questionable as a long-term goal; like ADSL, it too is highly assymetric, and it's really too expensive. (I think Verizon is doing it mainly for political show, and will slow down. Besides, FiOS is bundled with Verizon Online, with its onerous rules and likelihood of draconian censorship in the mid-term future.)
Still, I think it's premature to count out cable technology. Hybrid Fiber-Coax is an evolutionary path to bring optical fiber to the home. A decade ago, it was first being rolled out with maybe 1000 homes per node (optical transition node, where a strand of fiber turned to coax) and up to three analog coax amplifiers on the coax side. Modern builds have maybe 50-100 homes/node and no amplifiers. Thus far fewer users share the same capacity. DOCSIS 3.0, now being tested (CableLabs is very strict on compatibility certification), uses more than one 6 MHz TV channel at a time in order to boost download speeds. And while upstream is still a bottleneck, DOSCIS 2.0 tripled upstream efficiency over the original cable modems; as each DOCSIS 1.x modem is phased out, overall capacity can increase. There are also tricks for boosting upstream on a point basis by using the spectrum above 900 MHz as well as below 42 MHz, while cable companies can also just drop off fiber at a location that really needs it (not a house, but a business or multiple-dwelling-unit site).
Next glitch: The protocols themselves. TCP/IP is from the 1970s, and while it's amazing how far it's gotten, it is really not designed for today's applications. IPv6 is the wrong approach -- tastes crappy, more filling. We really need an all-new protocol stack; it's not obvious how to phase it in though, or get consensus on a replacement. Remember TCP/IP happened because the government financed it for its own internal use (ARPAnet) and Berkeley produced open source code for it, so it became a de facto standard for multivendor corporate networks too. (This during the 1980s when OSI was supposed to be the standard, and most companies used their vendors' proprietary network technologies like DECnet, IPX, SNA and Wangnet.)
Plus there's the business issue: It's hard to make money providing Internet service. The early public ISPs were subsidized by the 1990s stock bubble. Telco/cable duopolies are potentially profitable (actually, telcos may still be losing money at it, though cable does better) but pure ISPs have a tricky time meeting demand with the kind of prices people want. Since there is usually no price feedback, users have no incentive to not do things that cost their ISP a lot of money (streaming HDTV, lots of big DVD downloads, etc., especially from distant sources). ISPs prefer the proverbial little old lady who just uses the computer to check email and stock prices a few times a week.
Norway, Sweden, Canada, Finland have lower pop. density than the US and better, MUCH BETTER broadband.
This stupid argument has been debunked a zillion times, including a few times in this very page already.
The only reason why broadband sucks in the US is because of CORRUPTION. Legal corruption, but corruption nonetheless.