Deeper into on-net buildings

May 29th, 2008 by · 8 Comments

Yesterday I gathered on-net building data. Now, size isn’t everything, it matters a great deal what buildings you connect and how you connect them.  So let’s see what we can get out of those numbers.  In terms of on-net buildings per market served:

Company Buildings per market
TW Telecom 114.5
RCN Metro 100.0
Abovenet 86.7
XO Communications 78.9
Integra 72.5
Level 3 60.8
AmericanFiberSystems 60.0
Cogent 35.4
Zayo 8.9
US Signal 7.1

What does this mean? Maybe not much, but it’s interesting anyway.  I suspect the Zayo numbers are inaccurate though, Zayo doesn’t easily differentiate between regional and metro fiber, and thus their 90 markets includes many where rather than having an actual ring they have regional fiber just passing through as part of a larger loop – so the count of 90 markets may not be comparable to the others in the table. Some of the other differences come from the size of market, it might be harder to light up 100+ buildings in Roanoke VA than it is in New York NY where there are obviously more large buildings in less space and thus more buildings one can make a financial case to bring on-net. Level 3 has many smaller markets throughout the east, perhaps that brings down their numbers somewhat. RCN Metro and Abovenet concentrate on the very largest cities, which shows up perhaps in their density.  Above all though, the numbers show why TW Telecom is often seen as the leader of this segment, their portfolio of on-net buildings is both large *and* dense.


Another way to look at density is perhaps metro route miles per on-net building.  A lower number implies a higher usage of the fiber – if the average building is similar.  This is a measure of physical compactness of a company’s networks, I’ll neglect Zayo for now until I figure out how to handle their route miles number.

Company Route miles per building
American Fiber Systems 1.9
TW Telecom 3.0
XO Communications 3.0
Level 3 3.4
Abovenet 3.5
Integra 3.8
RCN Metro 4.7
US Signal 7
Cogent 8.4

And now we have American Fiber Systems at the top of the list, and that is no mean feat when one considers the markets they serve.  They are a private company that few people know of, let alone understand or look at, and there isn’t much public data on them.  Yet they not only survived the crash and the ensuing telecom nuclear winter, they did so without making any fuss at all, perhaps now we get an inkling as to why.


Now, I find the consistency of the next 5 numbers very interesting, because the companies themselves are so wildly different.  How is it that such different business models produce such similar looking statistics of 3-4 route miles of fiber per on-net building? It’s eerie, and I’d really like to figure out why.

RCN Metro’s relatively high number surprised me at first, given the compact nature of their east coast markets, but then I recalled that half of their business was the old Neon, which did not serve the enterprise directly and thus would cut down their density.  Likewise, US Signal is higher because they are the only one in the group that does not serve the enterprise directly in any substantial way, thus they connect fewer buildings.  I think Cogent’s position as the least compact on this chart may have to do with the fact that all their metro networks are entirely IRU based – they own relatively few fiber miles and no conduit miles (that I know of) – thus they have a different cost structure than the rest.

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: CLEC · Metro fiber

Join the Discussion!

8 Comments So Far

  • Frank A. Coluccio says:

    Rob, I’ve been enjoying this and the other threads you’ve started here. Great idea, and some great content. Thanks.

    I understand (or at least think I do) why you are not showing the incumbents’ stats. However, it might be interesting if you did insert AT&T’s (TCG’s) out-0f-area fiber properties and Verizon’s (MCI’s/MFS’s) out-of-area fiber properties in your charts. Maybe you could produce one chart with those, and one without, fwiw.

    Qwest, too, has fiber rings outside its own service area, although I believe it piggybacks many of those atop others, in the main. Which brings up a good point, since, I’d venture to guess, some proportion of the buildings you’re showing as lit — by Carriers C and D, say — are actually being delivered by virtue of piggybacking arrangements over Carriers A and B. Thoughts?


  • Rob Powell says:


    My problem is finding a source for that data, VZ/T/Q seem to consider their out-of-region assets as almost non-material and don’t tell us much about them, whereas for the providers I was looking at their on-net buildings are their lifeblood. Do you have a source for these ILEC out-of-region fiber assets and buildings?

    Regarding piggybacking, most of those I mentioned own most of their fiber, the rest via IRUs. They would not consider a building on-net if they paid another carrier for special access facilities. In the case of Cogent, it is all metro fiber IRUs, mostly from Abovenet and Level3 I believe, with their own laterals to buildings. But in terms of fiber ownership versus an IRU from another telecom, yes there would surely be overlap there in some cases.

    FWIW, much of Qwest’s out-of-region fiber rings came from OnFiber, which in turn came from Telseon and Sphera – and I think those two owned their own fiber though I’m not sure.


  • Frank A. Coluccio says:

    One afterthought, if I may. You noted:

    “How is it that such different business models produce such similar looking statistics of 3-4 route miles of fiber per on-net building? It’s eerie, and I’d really like to figure out why.”

    I’ve no direct evidence of this, but I suspect that my earlier point concerning piggybacking arrangements may explain at least a part of the tracking between A-models and B-models. This would be consistent with your observation, since facility-less fiberco business models differ from those that are purely facilities based.

    In a campus implementation in Jersey City I was intimately familiar with not so long ago, for example, four “facilities-based” carriers delivered a minimum of 72 strands apiece (one came in with 216), but two out of the six were using other peoples’ (who were also included in the six) glass. On top of these four, naturally, other LD carriers and resellers leased access, too, but the four I’m referring to here were classified as “facilities-based”.

    I later learned that this same donor-parasite relationship I had witnessed before me was also being played out by the same entities in other locations, as well.

    The situation I just described took place at the height (deepest part?) of the crash, so perhaps there were more incentives on everyone’s part back then to give and take than there is today.

  • Frank A. Coluccio says:

    We seem to have crossed each other in the ether. I wouldn’t be so sure about the claims of ownership as you seem to be. As for the AT&T and VZ properties in question, you may be able to dig those building counts (or even the actual addresses) out of the DoJ procedings that ended earlier this year (or last, not sure) settling the dual building coverage issue following the SBC-AT&T and VZ-MCI mergers.

  • Frank A. Coluccio says:

    As one might expect, the DoJ site did, at one time at least, contain the building maps and addresses associated with VZ’s (MCI’s) and AT&T’s (TCG’s) metro builds, but those have since been redacted:

    Also, after giving further thought to the fiber rentals and IRUs we discussed earlier (i.e., those that I cited as being “piggy-backed”), I’m now inclined to agree with you that those would not ordinarily qualify as buildings on a renting or leasing fiberco’s “system” or “build”, although, marketecture being what it is, they sometimes do appear on a provider’s list of “lit buildings”.

    As for IRUs, IMO, due to the sometimes confusion and unfamiliarity of many with this subject, it’s probably a good idea if you defined exactly what you mean by IRU, and if a taxonomy of IRUs is in order, then that would be a welcome addition, too.

    My earliest experience with IRUs in a telecommunications context came during the age of telegraph, when International Record Carriers’ (RCA Global, ITT Worldcom, WUI, etc.) customers had purchased “IRUs”, which were individual telegraph system channels rated at speeds as low as 75 baud.

    At about the same time, voice-frequency IRU channels were bought and sold on transatlantic cable systems that were nothing more than analog 3KHz AM/FDM’ed channels used for both voice and data. These were capable of supporting a blistering 2.4 kbps, at the time, if the channel was properly “conditioned” to C-4 CCITT channel specifications.

    Later on, I had occasion to negotiate what might be thought of today as the equivalent of an IRU (actually, several of them, with a CAP), during the mid Eighties, only they called them Optical Fiber Services (OFS Contracts), instead.

    Since the bubble years, some confusion has arisen concerning just what an IRU is. It has at times connoted the outright ownership of a fiber strand, cradle to death; or sometimes it specified indefeasible rights to use a strand for a specific term of 20 years, say; more recently it also has been used to mean indefeasible rights to use an individual wavelength.

    So it’s probably a good idea to define what we mean by IRU in this forum.

  • Rob Powell says:

    Well, in the context of my comment, I meant simply a long term metro fiber IRU, e.g. 20 years of usage of a number of fiber pairs with intermediate splice rights etc. But you are quite right, the term IRU spans many types of agreements from fiber on up to various data circuits. Even fiber IRUs vary widely I suppose, there are agreements out there specifying how much fiber on a network one has a right to, but allowing the routes to be chosen later – I think WilTel had one with Abovenet back in those days, I guess Level3 has it now…

    You are right that it is messy, and the industry is always loose with its marketing terms, ‘lit building’ being one of them. But that ambiguity hasn’t stopped me yet from trying! 🙂

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