OFS: What Fiber Glut?

So you’ve heard there's a fiber glut going on? Well you’ve heard wrong -- at least according to Optical Fiber Solutions (OFS), a designer, manufacturer, and supplier of fiber optic products, which hosted a breakfast at the Optical Fiber Communication Conference and Exhibit (OFC) yesterday to address the topic.

The company, which Lucent Technologies Inc. (NYSE: LU) sold to Furukawa Electric Co. Ltd. in a much-publicized deal last year (see Lucent Cuts Deal on Fiber Unit), announced an array of new products yesterday (see OFS Unleashes Fiber Flood). And, in its presentation, OFS not only tried to dispel what it called “rumors of a fiber glut,” it claimed there is far too little fiber out there to serve the enormous demand looming on the horizon.

The debate around whether or not there’s an overabundance of capacity available in carrier networks burst into flame last summer, when the results of a report by Merrill Lynch & Co. Inc. analyst Michael Ching were published in the Wall Street Journal. The report indicated that carriers use only about 2.7 percent of their total lit fiber capacity (see Fiber Utilization Figures Challenged).

But OFS said this report and others, along with a long line of articles, miss the point. “Telecom networks, like highways, must be sized to handle demands at peak periods,” said Janice Haber, OFS VP of systems engineering and market development. She pointed out that the "peak-to-average" rate of data networks is much higher than it is in voice-centric ones -- typically ten times higher, in order to accommodate high-traffic bursts.

Using the 10:1 peak-to-average rate that Haber said data-driven networks require, Merrill Lynch's estimate that 6.4 percent of capacity was being used at the end of 2001 translates into 64 percent peak usage. This, she said, is dangerously close to the 70 percent threshold beyond which customer service can no longer be guaranteed.

In addition, Haber pointed to a number of studies that show the growth in demand for bandwidth has remained steep and fairly stable for decades. One of the reports, issued by IDC last December, estimates Internet bandwidth growth at 147 percent.

OFS isn’t the only company calling for more, not less, fiber on the market. On an earnings call last month, Wendell P. Weeks, president of optical communications at Corning Inc. (NYSE: GLW), said that Corning believes the optical downturn is bottoming out and that carriers are at more than 55 percent of their wavelength utilization. Corning, he said, is planning new fiber builds to meet an inevitable demand (see Corning: 'We've Hit Bottom').

So why all the gloomy fiber forecasts? One problem, according to Haber, is that there are several different ways of measuring how much capacity is used, rendering the concept of “average capacity utilization” meaningless.

Some studies, for instance, measure the active capacity demand, which is the total bandwidth demand divided by the sum of capacity of active channels. Others measure channel slot utilization by dividing the number of active channel cards in data networking systems by the total number of slots installed. Yet other surveys divide the number of lit fibers by the total number of installed fibers to measure so-called fiber utilization.

A problem with measuring fiber utilization, according to Haber, is that stranded fiber, which will probably never be lit, is brought into the equation. “Not every fiber gets lit to terabit capacity,” she said. “That’s just not the way it works.”

The drop in fiber sales has nothing to do with network planners responding to a drop in demand, Haber asserted. Rather, it is a response to Wall Street's fears following the false "fiber glut" rumors.

“This thing has really frozen decision-making with service providers,” she said. “But it’s coming from fear more than from knowledge.”

Whether or not the fiber glut is just a figment of analysts’ and journalists’ imaginations, the fact remains that last quarter was the worst ever for fiber sales, and that after just one-and-a-half months of contributing to Furukawa's financials, OFS has posted a loss of about $77 million (see Did Furukawa Buy a Lucent Lemon?).

Having had her say, Haber remains optimistic. “My experience,” she said, “is that when this business goes down sharply, it comes back with a vengeance.”

— Eugénie Larson, Reporter, Light Reading
<<   <   Page 3 / 3
DoTheMath 12/4/2012 | 10:44:31 PM
re: OFS: What Fiber Glut? LightBeating: I just want to point out that with all the technology that we have now, it is really, really hard to understand why the customer cannot get those bits for cheap, as opposed to the Mips of my computer, or the Gigabytes of my DVDs. Hell, we can't get those bits at all in most places.

Couldn't agree more LB. I think one solution as to how to get innovation moving, and get more and more bandwidth every year for a given fee is to move to a user-owned-infrastructure model. In computing, end users own the PC, the ethernet cards and the switches. This effectively means that there is an "incremental adoption model" where affluent users in effect subsidize the R&D so the later adopters reap the benefits.

Here are some elements of a solution for the new network:
a) Municipally installed fiber to the home, with open colo facilities, open to any service provider paying a set (smallish) fee to the city. In California at least, cities have to call referendums before they raise the bond money to pay for fiber-to-the-home, and my guess is that most cities *can* get support from the public. This solves the right-of-way issue neatly, and may not cost more than about $2K per home, so with tax-free muni-bonds used to finance them, the property taxes may go up by $20/month (if that). In San Jose, I pay property taxes about 20 times that amount right now. We just passed bonds in San Jose to fund school renovation that are several times any fiber-to-the-home project would cost. In fact, schools may be given free colo space for beaming educational programs; politically this plays well with suburban constituents.

b) Open network regulation: if the FCC mandates that all network services offered on this municipal fiber have to be based on openly published interfaces (so that anyone could build end user equipment or software), this will stimulate competition and innovation on the end user box business, not to mention the infrastructure box business.

c) Competition can now be unleashed between the Bellhead service models vs the Nethead service models. I am partial and predict that Bellheads will be trounced before the game even starts, which is why RBOCs will lobby violently against this model.

In fact, enterprises may be beginning to bypass traditional service providers and leasing dark fiber or lambdas to run their own network services. This is happening for campus or intra-city networks first, and looks like it will spread.

I strongly believe that a "enterprise like" network model is possible in this industry. All we need is the courage to think out-of-the-RBOC-box. I have served the RBOC types for several years, and do not relish continuing this career, if they are the only game in town.
gea 12/4/2012 | 10:44:28 PM
re: OFS: What Fiber Glut? The recent posts are good examples of the big time misconceptions data-only guys have about how things work. I remember at a startup I was at when we were discussing "packet over SONET" a router guy actually said: "SONET? No. We don't want to do SONET. The future is data." I asked him how he thought data got from one part of the country to another, and he pretty much just thought there was a continuous string of routers and switches cross country joined by 10/100BaseT and GbE links.

And for anybody who thinks that by printing that statement I'm trying to say "The future is data and not SONET"...you really need to do some homework, and fast.

Data doesn't care if it's in a format a router guy is familiar with or not. Packet-over-OC-192c is just as much a "data" format as GbE or 10GbE (uh, which has SONET framing in two out of three of its formats). Even that article about special opps building out their infrastructure shows how absolutely blind LAN guys to how their data travels across town or cross country (hint: it doesn't remain in what you would call a native data format). All data CLECs (Yipes, Cogent, and others) already exist, but suprise! They map their data links over SONET, and use Cisco 15454s (no, that's not a router). Why? Do your homework and find out.

Now if you want to make the argument that circuit switching is holding us all back, well sorry. We all know that already. We also know that the RBOCs make the lion's share of their income off of voice (even though it's less than %50 of their networks). They don't have a lot of motivation to eliminate circuit switching or voice. So of course it's slow.

At the same time, however, nobody yet knows just HOW to get away from a circuit switched world. And oh yeah, I'm sorry but throwing giant pipes at the problem won't make the QoS issues go away. What will be needed is something like an MPLS-based network that allows for protection switching, priority-based billing, and subscription management. And then, even with the technology in place the real problem is in operations.

Oh yeah, and don't forget: that switch, router, and transport infrastructure's going to have to stay up for more than 30 days at a stretch, otherwise operations costs will eat your lunch. And it'll have to survive in 0C to 45 or 50C (sorry...a CO is not just one giant computer room). And what about lifeline services? Will this data network that's so cheap and easy have to support that? (Answer: Yes.)

So my point is, these religious statements about data get quite tiresome. I'll make a flippant statement that probably has more truth to it than any of us want to believe: if true MAN data networking were so obviously cost effective and easy to manage AND profitable, there'd be big comapnies making money at it already.

We all want it to happen, and we all know there are barriers. Oh yeah, we all know there are vested interests. But the best minds in the industry don't know how to get from here to there, and no simple set of easy answers is going to do it.
<<   <   Page 3 / 3