Bandwidth Speculation's Legacy: Dark Fiber

Darwin O'Connor pointed out this article in thestar.com which "says there has been massive amounts of fibre-optics put into the ground that hasn't been hooked up because of a lack of capital, despite Internet capacity problems. It compares the situation to the railways in the 1870s." It's tantalizing that there's so much bandwidth via fiber, but prices aren't exactly dropping for home users.

Comments

Dark fibre and margins

[anticypher]

There was a great article a month ago in the Economist about how telecom companies over-investing in fibre was the main cause of the recent economic blues. The general sense of the story was how a few buyouts of a few metropolitan fibre companies in 97-98 sparked a huge boom in investment in laying fibre.

But once the fibre started to be lit up, a dozen large telecom companies (nortel, alcatel, lucent, etc) started to compete with each other for cheap bits. When the price started falling, everyone realised there was too much capacity, and not enough margins reselling the bandwidth.

The equipment manufacturers (cisco, lucent, redback, etc) had sold tons of fibre terminating equipment to start-up and established telcos by financing long-term loans. Those loans were based on the (bad) assumption that prices for bits would stay at the same level. Prices dropped, and now many telcos don't have the income to pay off their loans. Since there is no more investment in new fibre termination equipment, the fibre will stay dark until the next economic boom.

There just isn't enough cheap hardware available to terminate all this dark fibre. There is literally tons of very expensive kit sitting in warehouses that cisco, nortel, and lucent can't sell. When those machines have buyers, then we will see prices continue downwards.

This is an economic problem, not a last mile problem. When the economy turns back up, then that fibre will start to light up as well, and long-haul prices will stay low, but bandwidth demands will increase.

the AC

Whining/fibre are cheap, support/routers cost $$$$

[eschasi]

This is an old story, and one that makes people immediately leap to the wrong conclusions. Executive summary: fibre is the easy, cheap part. You want fibre to your home? Sure, I can do that cheap. But if you want to connect it to something else at fibre speeds, well, that's another story.

I live in a small town in Michigan (Dexter, 12 scenic miles from Hell). We have a major Sprint node in town. A civic group of local civic technologists would love to tap into that, getting our tiny town some huge bandwidth. We love living here, and would like to see the internet connectivity being fast and cheap enough to attract Internet-based businesses.

But we can't currently generate enough $$$ to make it worthwhile. What Sprint has in place is basicly an optical repeater. To expand it into data service, they'd have to install an OC-XX router, where XX is probably greater than 12.

Ever priced an OC-12 router? Or facilities to support it remotely? The DS-3 cards and T1 cards to step down those speeds? The lines from your home or office to that router? The maintenance agreement on all that equipment?

But everyone would use it if it were available you say? Show me. Figure out that cost, go talk to your neighbors. See what you can get the cost per connection down to. We went thru that exercise for getting gas lines installed out here. Even with a three year payback, people weren't interested in anything over about $1000 one time cost. I'd be shocked if your neighborhood was willing to go $1000/ea plus $100/mo for ethernet to every home.

As an example proof of fibre being the easy part, I believe that WilTel some years ago sold their entire fibre infrastructure to some other company for quite a nice price. They retained one (count it) one strand in each bundle. That's been enough to handle all their traffic. It was brilliant, just frigging brilliant. Now someone else has to pay all their fibre infrastructure costs, and they just get one card in someone elses equipment.

dark fiber?

[tewwetruggur]

Isn't that for constipated evil people?

That whole last mile thing...

[baptiste]

There is no doubt that there is enough backbone fibre in teh ground (once lit up of course) to handle magnitudes more traffic. Remember, while fibre was getting cheaper to put in the ground, companies like NORTEL were fitting up to 160 (yes one hundred and sixty) channels of data on a single fibre by splitting and recombinign different frequencies of light.

This tended to make the dark fibre problem worse in that less fibre was needed to carry more data.

The whole problme is the last mile to the house. Until there is an economical way to get high speed data to the home, its never going to happen. Cable modems stink bacuse of shared bandwidth (once you approach saturation - look out) but they rock now because they go EVERYWHERE. DSL rocks if you're near a central office - and with speeds in development approaching 4MBps and higher - sweet!

But DSLs problem is the need to be close to a CO. TO serve neghiborhoods, the telcos have to run fibre to each neighbor hood and concentrate the DSL connections there - cable has to do something similar - concentrate and backhaul to the data center. But runnign fibre to neighborhoods is EXPENSIVE!

I have to admit though the whole dark fibre phenom is encouraging. I watched AT&T drop new fibre along a right of way that runs through my nextdoor neighbors pasture. They dropped SIX fibre conduit in the ground - each capable of holding a LOT of fibre. Sticking it int eh ground is the labor intensive part - now that the conduit is there, if their fibre needs GROW beyond whats already IN the ground, they just snake mroe fibre through the unused conduit.

Even better - once they main fibre run was in teh ground, two more conduit were run alogn our road straight to - the local telco. Very encouraging :) Provides them with MAJOR bandwidth connectivity in teh near future (which believe it or not, they already have - they're a SMALL telco with an auxillary fibre backbone busienss go figure)

But the bottom line is, until teh last mile problem is solved in a fashion that can make the telcos/cable comanies/whoever some decent money that we're willing to pay - the bulk of that fibre will stay dark for some time.

more like highways than railways

[Proud Geek]

You can never have too much bandwidth. The situation is more comparable to post world war II United States, where there was a massive government program to build highways.

Initially, it looked really silly, since there weren't many cars and there weren't many people driving all over. Now the highways are full. Same with fiber; traffic will grow to fill them to capacity. It will just take a few years.

I'll leave it up to other people to decide if that's a good thing.

Re:Dark fiber glut? Duh!

[dachshund]

Each telecom laid plenty of extra dark fiber, to be sure, but they also assumed that a much larger portion of what they laid would be utilized, and this would pay the debt created by the build-out, in addition to the maintenance. Unfortunately, they assumed wrong, partly due to the fact that all of the telcos built their networks at the same time. Now they don't have enough money to pay for the massive last-mile upgrades that will make all of that fiber worth something, so it'll sit dark for much longer than it was supposed to.

Dark fiber glut? Duh!

[GuyZero]

Since it costs about the same amount of money to lay one strand of fiber as it does to lay 10,000 strands, it seems somewhat obvious that immediately after a big fiber laying expansion that the amount of unused capacity would grossly exceed the amount used.

I mean, if all those fiber laying companies had laid just barely enough to meet current needs they'd have to go back and dig trenches again to create more bandwidth. That would be expensive. And dumb.

The situation is somewhat akin to looking at a newborn baby and saying "He's less than 5% of the size of a real man!" A-duh. Check back in 20 years and see how much of the fiber is still dark.