Dark Fiber and the Law of Unintended Consequences

May 25th, 2010 by · 20 Comments

Allied Fiber’s formal announcement on Monday of its national dark fiber buildout rankled more than a few readers.  The conventional wisdom has been that following the dot com crash, no new national intercity conduit level network builds were even remotely feasible.  Prices had fallen too far, rights of way were too contentious, too much fiber remained untapped in the ground between major markets, and financial backers would simply never go for it.  Whether Allied Fiber succeeds or not in its venture, they may have already blown a fatal hole in that wisdom.  There are many factors that have contributed to this, but for a moment I’d like to focus on how the industry’s view of intercity dark fiber in the USA has changed, and may be changing again.

Back in the last bubble as the fiber barons built out their networks, selling or swapping dark fiber was a way to help fund the process and staggering amounts changed hands.  Literally dozens of domestic ventures were built around such assets, just about every European PTT held dark fiber IRUs across the USA as well, and far too many networks were lit.  Just a few years later, of course, carnage ensued.  Much of the new fiber got consolidated into a half dozen or so surviving intercity networks.  A new view of selling dark fiber emerged:  that selling too much of it enabled the competition and was therefore a bad idea.  You had to be careful about who you sold it to, where you sold it, and what other products could tag along to make it worthwhile.  Such newfound self discipline made so much sense in light of the experiences in the telecom nuclear winter that it quickly became the rule across the industry.

But that self policing has now run into the law of unintended consequences.  If customers want to buy dark fiber IRUs, and existing owners are reluctant sell them, then there is a market opportunity for someone to step into the breach.  Enter Allied Fiber with its innovative wholesale dark-fiber-only business model, a modern approach to on-ramps, a little luck from the timing of the low latency movement, some PTT customers in the wings – and lo and behold the initial funding wasn’t actually that hard to get.  One of the major selling points Allied Fiber has used is that there is a shortage, not in the quantity of fiber in existence but in the quantity of fiber actually for sale via dark fiber IRUs since existing providers have given such priority to lit services.  The decision by the industry that made so much sense now risks being turned on its head: instead of less dark fiber there may now be even more.

But Allied Fiber may face its own unintended consequences.  If they succeed in breaking the dark fiber logjam on the major intercity routes, the Level3s and Qwests and AT&Ts holding their own dark fiber in reserve could now be forced back into the market, thus driving prices down again.  After all, someone’s going to sell the stuff, it might as well be them.  Whoever sells the fiber, though, the fear is that a plethora of dark-fiber-enabled networks will bring back the dreaded buzzphrase ‘fiber glut’ in spades.  

Will that happen?  I suppose that depends on what people think they can do with intercity dark fiber now.  Ten years ago, they all thought they could sell the same stuff to the same people and exponential traffic growth would take care of everything.  But it didn’t work out that way and each learned the hard way just how much it costs to run your own network from the fiber on up.  Even the European PTTs mostly returned their dark fiber IRUs rather than continue lighting it: FT, DT, C&W all did in one form or another.  If they return again, surely it will be with a different plan in mind?  Is there an ecosystem in which hundreds of intercity networks at the dark fiber level can coexist?

If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!

Categories: Fiber optic cable · Internet Backbones

Join the Discussion!

20 Comments So Far


  • Anonymous says:

    Its only on low latency paths, and in this case its only Chicago to NYC, maybe DC. Take away low latency, there is no investment. And Allied is not the fastest, so their multiples are shrinking and will continue as existing providers optimize and can offer customers more support, products, etc.

  • carlk says:

    Excellent essay for describing the things that continue to swarm like bees in my head whilst this abomination-DF supply increases-was announced. Thank you for your wisdom as well as providing historical context.

    If there’s any good news, it’s that persons with names like “anonymous and anon” are seemingly seeing ghosts of Christmas past pointing to extreme pressure on titans turned dinosaur once again.

    I’d like to know more about who is building this for them, and who they’re going to lease facilities from?

    I won’t be surprised if and when it is more thoroughly disclosed or explained, that it involves (3) and/or Kiewit somehow.

    It’s the bottlenecks in the last mile stupids, and the bottlenecks must be BROKEN so that internet traffic can spread to the backbone like WILD FIRE!

  • Al B says:

    couple items on your great article concerning the latest NEW fiber being installed:
    1. will LVLT now stop preaching that they have the newest in-ground fiber?
    2, seems as if it was only a short time past when LVLT was selling DF to companies like GOOG etc and no announcement was ever made as to how thye were going to use this DF – or have I probably missed what’s going on in the new world??

    • Rob Powell says:

      1. LVLT preaches that it has the newest *national* fiber in the ground. Until Allied or some other network is done, they still do.
      2. There have of course been some dark fiber sales – just careful ones, and notice that none of them have turned into wholesale competitors in a long, long time. For that we have XO, but those IRUs were sold in 98 or 99.

      • Steve Maloney says:

        Has DarkStrand, the for profit arm of NLR, made any sales of dark fiber of note?

        How do you think they will ft into this market?

        • Frank A. Coluccio says:

          Steve, the name dark strand is, in my opinion, a misnomer. True, they use strands that were initially acquired and redeployed, unsurprisingly, in dark form (in reality, strands that NLR acquired), but their service offerings are over lighted spans, hence they are not dark, per se.

          I’m led to believe that, at a minimum, they’ll deliver a wavelength, but more likely some denomination or multiple of gigabit Ethernet. See the video below, whose sub-heading (directly beneath the screen) reads “Fully-Managed Lit Fiber Service”:

          http://www.youtube.com/watch?v=QJBfHPYMg-Y

          That said, they may well support dark fibers as well, but as one-offs, not as their mainstay however. Also worth mentioning, DS initially set out to offer (and as far as I know, still does) high-performance computing “platforms” based on SOA-like architectures to corporate divisions that fulfill scientific and research missions comparable to that of academic and research laboratories. There must have been mission creep since those earlier days, though. In one early publication they espoused a DarkStrand “ecosystem” in support of research-commerce partnerships. My guess is that their strong suit is software and application development and research activities, more so than day-to-day production, but I could be wrong.

          frank@fttx.org

          ——

  • Frank A. Coluccio says:

    Hi Rob. I join the other board members here in applauding you for an inspiring and well-written essay. Thanks.

    Second guessing this sector very often proves futile when viewed retrospectively. Several years ago the submarine cable group vowed never to repeat the sins of the past (read the editorials and side stories of Subtel Forum Magazine since 2002, e.g.), only to follow through with a torrent of routes during the past two years that is delivering orders of magnitude more capacity than anyone ever imagined. Africa and PACRIM stand out in this respect, which placements are now proving to have more than a mere ripple effect everywhere else throughout the world.

    Greed trumps reason, technology stymies our learned beliefs, and the bandwidth ultimately gets consumed no matter what. Someone once argued that slot machines in large casinos required only RS-232/485 asynchronous connections to the data center for the amount of bandwidth they consumed, and that wiring the entire casino with fiber would be foolhardy. Until it was asked whether a slot machine could pull double duty and support an Internet kiosk function, or, if retinal scanners could be incorporated into each unit along with a live video feed back to the casino’s security office.

    Doubting the propensity of new applications to consume previously unfathomable amounts of bandwidth in the most unexpected and often seemingly bizarre ways has long been a game played by fools and hedgers. Only those who cherish the fruits of scarcity have the filters to see this differently, fwiw.

    As for the ‘newest in the ground’, I’ve got to chuckle sometimes, recall the great build-outs of ten years ago when three different, very-large fiber manufacturers were pitting their brands and unique claims of capabilities against one another at a time when the looming effects of polarization mode dispersion on transmission speeds that exceeded 10 Gbps (actually, 10 gigabaud, since symbol rate matters more than derived bit rate) were only first becoming understood.

    So, today we are seeing 40Gbps and 100Gbps capabilities rolling out of the gate, but they are using multi-level phase and amplitude modulation schemes based on 10Gbaud, not 40 gigabaud and 100 gigabaud, respectively, for which (latter baud rates) much of the transmission improvements in later day cable designs were supposed to accommodate.

    Also consider, we now see signs that maverick-like modulation schemes capable of using orthogonal frequency division multiplexing may be used to achieve super 100Gbps rates going forward using existing, even aging, SMF that was placed in the ground twenty years ago.

    Unanticipated developments, unintended consequences, indeed.

    frank@fttx.org

    • Frank A. Coluccio says:

      Here’s a good article, which is far less flippant, admittedly, than my earlier allusion to using aging SMF plant, from Gazettabyte:

      Optical transmission beyond 100Gbps
      By Roy Rubenstein | May 14, 2010
      Briefing: High-speed optical transmission.
      Part 3: What’s next?
      http://tinyurl.com/3262kta

      Part 1 of this series delves into the OFDM approach to 100Gbps and beyond, which I also mentioned, here:

      Ofidium to enter 100Gbps module market using OFDM
      DateMonday, April 5, 2010 at 2:42PM
      http://tinyurl.com/y3753qz

      Enjoy,

      frank@fttx.org

      ——

  • Dave Rusin says:

    A wholesale, dark-fiber only, IRU/Lease model comes with little overhead than those more vertically inclined … it will make money … remember in DF, relocates and maintenance are pro rata to those using the network … very low overhead needed and opex once built.

    OPEX needed — to light a few strands like a NOC to monitor for outage, micro bending,etc. … not a big expense.

    Building it where they are not is important as well ..

  • Anonymous says:

    Is the opposite true? Does the last mile bottlenecks give opportunity to new entries into that market who could compete with existing providers even though they may be edging out their own network? And could these newco’s piece together their own national network again as the DF market loosens back up?

  • carlk says:

    Seems plausible if you believe “time to market” while new entrants are talking about huge “fiber demand” for embarking upon such a plan, is measured in a DECADE or more again?

  • Anonymous says:

    I think the real value Allied Fiber could provide on their dark fiber routes is if they can support increased spacing on the regen sites on their long haul routes. Increased spacing allows fewer electroincs along the route, which equals lower cost structure to provide for lit services. This is the perceived adavantage that Level 3 has always touted. If there is a newer generation of fiber availble that allows this, and Allied can better the span distance for regen that what Level 3 and other regional fiber providers support, then there may be a market here.

    • vferrari says:

      With the advances in DWDM systems discussed earlier, wider spacing could be perceived as a negative. It used to be that fewer amplifiers to deploy was significant but that’s when amps were expensive. Today they are relatively cheap. The wider spaced network incurs greater loss per span. Overall reach disadvantage results in more frequent regen not less.

    • Frank A. Coluccio says:

      The regenerator hut spacings jumped off my screen as well. Your points are to some degree valid, since newer generations of glass can in fact sustain greater distances at higher capacities. This could arguably be seen as the basis of the Uganda vs. Huawei overland cable placement dispute of several weeks ago. See: http://tinyurl.com/37y2rhv

      However, even newer and more capable generations of glass will not be able to “save you” when future modulation schemes supporting signaling rates that extend into the multi-terabit range begin to appear. I also think the closer spacings are more conducive to the company’s latency objectives and their perceived ability to aggregate at a more local level than 120 km spacings and greater would allow. These are merely my own observations, the company may have other reasons.

      frank@fttx.org

  • The_highwayman says:

    I Did some boots on the ground recon and talked with a buddy at American Tower corp and he knew exactly who Allied fiber was and what they are doing…he never came out and said they were partners and I was intentionally vague at first…he also mentioned crown castle was pretty excited over this as well….

    take this info FWIW, is was not bona-fide concrete info, it’s recon only…

  • Dave Rusin says:

    The model is a very good model where Allied is building, what they are building and wholesale back-haul. Let’s throw in some interconnection business on top of it all.

    Low overhead, not boring through concrete or granite …

    The regen/latency worry-warts — the new pump lasers will take care of that …

    Allied is also cream skimming the copper RLECs as they build … whether towers, MTSO or Enterprise.

  • carlk says:

    Painfully aware of the short comings for relying upon what Jim Crowe says, versus what he does in the marketplace, I am keenly aware of his reluctance to be ALL THINGS to ALL tower owners with respect to “back hauling.”

    He has made it very clear how “investment capital” must be viewed when building connections to these giant radio antennas scattered across the U.S. landscape.

    This being said, considering 5 of the top five wireless carriers are (3) customers, even knowing that Alltel having previously being swallowed by VZ had caused some revenue losses, I find a new entrant attempting to ultimately go where no man has gone before, to be good for (3) fools in the bigger picture and final analysis.

    I didn’t call any of my “buddies” though! IMO

    • carlk says:

      With about 50% of the nation’s mobile switching centers on its fiber network, Level 3 is well-positioned to also connect base stations, but extending fiber to cell sites from there is an extremely expensive proposition, DeLong said.

      “Today most cell sites count demand in terms of number of T-1s, and they need about three to four,” DeLong said. “At about $250 each, that’s about $1000 in revenue. To build fiber to that location will cost $50,000 to $100,000, assuming your network is within 500 to 1000 feet of the tower. If you do the math — $12,000 a year – that’s a very long payback.”

      http://connectedplanetonline.com/ctia/news/level-3-backhaul-0401/

  • anon says:

    the top three wireless carriers are att, vzw and sprint. each owns a long haul network. plus, qwest has a deal with vzw. so, basically, any time a wireless carrier pays L3 to “backhaul” traffic from a tower to their LH network, they are doing so simply because L3 is cheaper and/or willing to invest the capital. strictly math. not sticky, easily replaced in 3 years or end of term. not an enterprise client, not differentiated. just cheap wholesale biz from a rival that presumably would rather put $$$ into new ringtones, phone subsidies, map ads, etc.

  • carlk says:

    You’re too picky, to understand STICKY. Soon, however, you’re going to a “STICKY BOMB” from (3) with or without James Q. Crowe.

    “Level 3’s CDN is highly scalable, meaning we’re able to help provide a fast, reliable, high-quality viewing experience for end-users, even if the audience watching online or via iPhone is massive (as it can be for major events),” said James Heard, president of European Markets for Level 3. “From the time the first serve is captured on video to the moment the viewer watched it on their computer or mobile device, Level 3 provided advanced technology and reliability to create a great experience, both for the customer and the viewer.”

    http://www.level3.com/index.cfm?pageID=491&PR=889

Leave a Comment

You may Log In to post a comment, or fill in the form to post anonymously.





  • Ramblings’ Jobs

    Post a Job - Just $99/30days
  • Event Calendar