• Featured Articles

100Gb Optical Networks – Getting the Timing Right

September 12th, 2012 by · 18 Comments

This Industry Viewpoint was authored by Paul Savill, SVP Core Product Management and Andrew Dugan, SVP Network Architecture & Engineering, both of Level 3 Communications.  You can find more of their thoughts on Level 3′s blog Beyond Bandwidth.

This year, Level 3 delivered a 15,000-mile, nationwide 100Gb private network for a customer and, recently, we’ve started to see other carriers announce their 100Gb network deployments. So, have we finally reached the cross-over to broad adoption of 100Gb? Maybe not just yet. Why? Because of why carriers are deploying it.

There are three main reasons why a carrier would deploy 100Gb today: 1) to lower its internal transport cost structure, 2) to fulfill a market demand for 100Gb wave services, and 3) because it is fiber-constrained and needs the extra capacity that the latest 100Gb systems can squeeze out of a fiber pair. Let’s examine these one at a time.

1. To Lower Its Internal Transport Cost Structure

The ability to lower internal transport cost structures is typically the driving reason behind the first deployments of new optical transmission technology. Usually, the big carriers themselves are also the ones on the leading edge of needing high capacity systems. But high capacity doesn’t always mean highest bit-rate of individual data streams or the lowest cost per bit transmitted. The problem for 100Gb technology is that while its cost structure is getting better, so is the cost structure for 10Gb. So 100Gb is chasing after a technology that is also continually declining in cost. Eventually, it makes sense that 100Gb will catch up and cross-over in cost effectiveness, but that chase for 100Gb is not yet over. Level 3 is still consistently seeing that on a unit cost basis we can get better pricing for 10Gb technology than we can for 100Gb technology. We hope that the crossover will occur soon and are continually looking for the right timing but the market has not gotten there yet.

Now, I know what you’re thinking – there is more to a network’s cost than just the capital cost per bit – there is also the operational cost. If your company is managing hundreds of 10Gb circuits running all over the place, there is an operational benefit to consolidating those into much smaller counts of 100Gb circuits. While this is definitely true, any meaningful value of that simplification can only be realized by very, very, large consumers of bandwidth. Level 3, for instance, has some core router city pairs in its network with more than 50 10Gb waves connecting them, but the economics including the operational cost assessments still don’t prove in for 100Gb at its current price point. You also have to remember that those 100Gb waves coming off the transmission system have to connect to something. If they are connecting routers, you have to consider the comparative cost of the 100Gb router blades as well.

2. To Fulfill a Market Demand for 100Gb Wave Services

The second reason to deploy 100Gb is to meet customer demand. While Level 3 has already turned up and delivered over 15,000 100Gb route miles of capacity to customers over the last year, most of that has been for highly specialized R&E and Government applications. So far, Level 3 has not seen a broad demand for 100Gb wave services in the carrier or enterprise markets. We are currently in the process of turning up additional customers with 100Gb services and expect to do more throughout this year, but those are largely circuits where customers are experimenting with the technology.

We have tens of thousands of route miles of deployed systems that are capable and ready to carry 100Gb services, but exactly when a new technology like 100Gb will start broadly selling can be tricky to predict, especially when 40Gb, which has been generally available for a few years now, has had such lackluster customer adoption. I think the main reason 40Gb never took off like 10Gb isn’t because the industry doesn’t need that speed, but rather because 40Gb all-in router and transport customer economics never beat 10Gb economics, and the 4-to-1 efficiency ratio was not wide enough to overcome the price disadvantage. Also, 40Gb interfaces never achieved the worldwide manufacturing scale needed to help it close that cost gap with 10Gb. But I believe 100Gb has more going for it than 40Gb. 100Gb is enough of a bandwidth jump to make a meaningful dent in operational efficiency from companies trying to consolidate assets, and the jump in scale is truly big enough to offer sustainable long-term cost efficiencies over 10Gb once manufacturing numbers reach cross-over.

3. It Is Fiber-constrained and Needs the Extra Capacity

The third reason for carriers to deploy 100Gb technology is to maximize capacity in situations where they are fiber constrained and are willing to pay a cost premium to squeeze additional capacity out of that constrained resource. The capacity that you can get out of a fiber depends upon a lot of factors, but if you assume you can get approximately the same number of channels out of a 40Gb system and a 100Gb system, that 2.5x increase could be valuable capacity to have even if it costs more. Although this could be a significant motivator for some carriers, it is unlikely to be one for Level 3 since we have over 165,000 route miles of fiber network with a relatively high fiber count.

So, in summary, it really boils down to cost efficiencies and market demand. Level 3 has the technical capability to turn up 100Gb circuits most anywhere on it’s nationwide footprint today, and in fact, we’re doing just that for specific customer cases. But it doesn’t make a lot of economic sense for us to productize a nationwide offering just yet because 10Gb economics are still so attractive and continuing to improve. We’ve found the market demand to still be limited to a pretty narrow set of customers. But it’s definitely emerging, and I have little doubt that in 2013–2014, 100Gb demand on a broad transactional wave level will be taking off – just not quite today.

Categories: Fiber Networks · Industry Viewpoint · Telecom Equipment

Join the Discussion!

18 Comments So Far


  • Anonymous says:

    Would a possible 4th reason be to drive bandwidth growth, maybe the google example?
    If you have the ability to deliver the higher speeds, wouldn’t it be in your best interest to differentiate your product with a disruptive style product?
    Just asking….

    • Paul Savill says:

      The issue isn’t as much about an ability to deliver 100Gb, it’s about delivering 100Gb at a price point that makes sense to customers. Certainly, the increased scale will support the next generation of bandwidth-hungry applications, but it will need to proliferate significantly first. Attempting to accelerate that proliferation with disruptive pricing that is at or below cost isn’t sustainable.

      • Anonymous says:

        Is there a set price point to 100gb? Is everything at this level an ICB? If someone is able to sell at a price point that makes sense to the first customer, do they maintain that price point to the next customer on the network or do they lower price for the next customer(s) until it reaches a market-driven point?
        Thanks for the perspective, very informative.

        • Paul Savill says:

          100Gb waves are still too new for a market rate to be established, and Level 3 hasn’t established standardized pricing for 100Gb waves at this time. But I can tell you that we will not establish a pricing scheme that keeps lowering the price with each new customer/circuit sold until the market rate is established. It will be ICB until we launch a standardized 100Gb wave product across a meaningfully large geographic region.

          • SamIAm says:

            “we will not establish a pricing scheme that keeps lowering the price with each new customer/circuit sold”

            That hasn’t seemed to stop Level 3 on 10G….

            • Parkite says:

              Agreed. It would be nice to hold the line on pricing for the health of the industry. Alas, wishful thinking. The kind of reckless pricing behavior from LVLT (historically) is how one accumulates a $12B retained loss since inception.

              • CoCo says:

                And yet, since there have never been any consequences to their dismally poor track record of earning a profit, why should they stop? 10G’s for $1000, I say!!! Damn the torpedoes!

    • Anonymous says:

      This Level 3 post sounds like Level 3 is a little bitter that XO was first in the 100G race.

      • beetlejuice says:

        XO was not first, and despite my dislike of L3, they are far better positioned to deploy it usefully on a mass scale than XO long term. All they need to do is get some more bottomless investor pits and BOOM – 100G nationwide, supported by a base thats far more embedded and growing than XO, and with a support model that isn’t schizoid.

        I still say the XO announcement is mostly treading water for a good buyer than it is seriousness about expansion of their reach, assets, and base. There’s simply no way Icahn has decided he wants to invest in the future of communications in America. He wants to invest in an image that gets the best price the fastest.

      • Anonymous says:

        US Signal was actually the first company to test 100Gbps across the USA and worked directly with Cisco on that.

  • anon says:

    “You also have to remember that those 100Gb waves coming off the transmission system have to connect to something. If they are connecting routers, you have to consider the comparative cost of the 100Gb router blades as well.”

    What about 10×10 muxponders where you would not need to change the client interface but you get the added operational value of having to manage fewer line-side links?

    • Andrew Dugan says:

      Yes, you could pick up a portion of the operational savings by using the 10×10 muxponders, and I suspect some carriers will, in fact, use this approach. But we’ve run that scenario at Level 3 and it’s still not enough of a savings for us to implement in this way.

  • CarlK says:

    For established pricing across the product portfolio of the Level 3 Network, at about the 28 minute mark, Ana Goshko, a B of A analyst, asks Level 3′s COO, Jeff Storey, about the health of pricing.

    This Dr. of Telecom is quite adamant that, “price decreases” in light of silicon economics–increased unit demand outstripping lower prices by creating more aggregate dollars– is quite POSITIVE for Level 3.

    Great “Industry Viewpoints” and personal commentaries expanding upon the Level 3 business model, by those Level 3 executives. Thank you.

    http://www.veracast.com/baml/media2012/main/player.cfm?eventName=1042_level_

  • Walter Scott says:

    When America is ready to take it’s training wheels off and scale, by flipping the switch, Level 3 will provide you with all the 100Gb waves you’ll ever need. Use the 100% IP Backbone of Level 3 to connect around the globe, to deliver content flawlessly.

    • Anonymous says:

      Level 3 & Flawlessly – Baw ha ha ha ha – rotflmao. Level 3 is a joke. There is only one reason why Level 3 hasn’t pulled the trigger – there debt went from $8.5B to $12B while net income decreased $140M. Double Whammy!

  • Like Mobile says:

    Drive bandwidth up maybe the only example

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
    Jobs by Simply Hired
  • Event Calendar