Enterprises: More Fiber in the Diet?

Large enterprise companies are increasingly looking at building their own fiber optic telecom networks, according to multiple sources throughout the industry. For a variety of reasons, it's become more attractive for a company to become its own service provider.

The long-term reduction in the cost of technology and the abundance of cheap fiber are driving the trend. Standards-based Ethernet and DWDM equipment is also part of the equation. Couple that with an explosion in VOIP, and you have a nice list of reasons why enterprises would consider investing in a new network.

Google (Nasdaq: GOOG) is the most dramatic example, but the trend is spreading -- albeit slowly -- to other large companies. For example, one source says Wal-Mart is building its own Layer 1 and Layer 2 network. Wal-Mart officials declined to comment for this article. (See Google's Own Private Internet.)

Are vendors taking notice? Consider the (NYSE/Toronto: NT) reorganization announced Friday. Nortel's new structure splits products into two categories: "enterprise solutions and packet networks" and "mobility and converged core networks." It's unclear what the changes mean, and the company won't publicly discuss the reorganization yet, a spokeswoman says. Observers pinged by Light Reading guess that Nortel wants more emphasis placed on wireless, the strongest card in its hand lately.

But the enterprise market could be a bigger catalyst for Nortel's plan. "Convergence is here and now, and our enterprise and carrier customers are demanding partners who can deliver enterprise innovation on carrier-grade platforms," said Bill Owens, Nortel CEO, in a prepared statement. (See Nortel Reorganizes Again .)

Examples of company-owned networks can be found among trading desks, banks, healthcare, and some cases of "good old-fashioned enterprises," says Chris Crosby, vice president of sales and technical services for Digital Realty Trust Inc., a firm that leases space for housing network equipment.

Even the (Nasdaq: CSCO) CRS-1, a box crafted with an obsession for carrier requirements, might be finding some enterprise customers. One source says that several corporate customers are toying with CRS-1s. A Cisco spokesman says he is unaware of any enterprise sales for the box, however. (See Cisco Stumps for CRS-1 and Cisco Unveils the HFR.)

Pouring it on
The idea of a corporation owning its own network isn't new. But some that haven't owned networks are taking matters in-house, partly because their networking needs are increasing so fast thanks to grid computing, on-demand applications, adaptive computing, and other new applications.

"Bandwidth need is increasing much more rapidly than people had anticipated, and the carriers haven't kept up," says Frank Dzubeck, president of Communications Network Architects. "The RBOCs are focused on consumers. They're not focused on enterprises."

"The amount of capacity they're driving in their own private networks has gone up so dramatically in the last couple of years, it's no longer cost-effective for them to lease their capacity. We're talking multiple 10-Gbit/s pipes," says Harmeet Singh, CEO of amplifier vendor Optovia Corp.

Of course, money is a great motivator for building your own network, as companies are always keen to cut down on their telecom bills. "It is still cheaper to get dark fiber than it is to ask your service provider to connect locations for you," Dzubeck says. "They're building it, not because they want to be their own service provider, but out of the unfortunate necessity of economics."

There's also the fact that equipment prices have gone down while some budgets have gone up.

"Capital is a little more free now, and big companies like to be in control of their own destiny," says Bernie McElroy, VP of business development at ADVA Optical Networking (Frankfurt: ADV) McElroy adds that the risk of owning and maintaining a network has shrunk in some cases, because optical networking experience has found its way into big companies. For that, thank the telecom layoffs of the past several years.

"You're seeing a lot more telecom guys in the IT department, and they say, 'Hey, we can do this ourselves,' " Crosby says.

Security is another factor, as many financial institutions need tighter control over their data. As well, governments and some companies don't want their network layouts to be known: By building the networks themselves, they can keep the locations of amplifier huts secret, Optovia's Singh says.

"Put it all together and some companies find that it is no longer worth the premium for all the risk they thought they were avoiding by leaning on a service provider," McElroy says.

Splitting the bill
Carriers don't necessarily lose out when a company considers building its own network, as they already offer managed services for those cases. Moreover, network ownership isn't for everyone.

In one instance, Banc One and Marriott International Inc. drafted a request for proposals (RFP) for a Wavelength Division Multiplexing (WDM) network that (NYSE: T) would manage, according to sources claiming to have seen the RFP. The companies planned to share the costs of building the network, with data for each company residing on its own virtual ring. AT&T would manage the network and provide the fiber. But even as those details are being passed around by equipment vendors, a Marriott spokeswoman says no such network is being built.

It might be early, then, to call do-it-yourself (DIY) networks a widespread trend. It encompasses only a few dozen large enterprises, maybe 10 of which are building enormous networks, Singh says.

"You're talking very large-scale users on certain specific applications," Crosby says. "I would be surprised if it's on an extremely large-scale basis. Actual network replications, you're probably not going to see, because the assets that are out there can be pretty well used."

Still, even the enterprises that don't build large-scale networks can be in the market for some carrier-grade equipment.

"What we're starting to see is enterprises wanting to aggregate traffic to get rid of local loops," Crosby says. In these cases, Digital Realty houses an aggregation box that's connected to the enterprise's data center. That box could be collecting traffic from the Multiprotocol Label Switching (MPLS) network, from voice and data networks, and from widespread point-of-sale locations, for instance. This allows the enterprise to avoid paying for all the local-loop connections that would otherwise bring this traffic home.

Even if the DIY trend isn't permanent -- few say fiber will be cheap forever -- vendors might view the enterprise market in general as a growth alternative for service providers. The carrier market has remained disappointing even in recovery, and it's not helping that the large U.S. carriers are gobbling one another up. (See SBC to Buy AT&T for $16B, SBC/AT&T: How Painful for Vendors?, and Qwest Qwits .)

— Craig Matsumoto, Senior Editor, and Phil Harvey, News Editor, Light Reading

fiber_r_us 12/5/2012 | 2:59:04 AM
re: Enterprises: More Fiber in the Diet? The Research and Education guys have already done this at a National (and even International) scale including metro and long-haul plants. Bill St. Arnaud and the various projects that community runs should be able to give you all kind of info on it. Other than their steep discounts from vendors (grants) and their willingness to be on the bleeding edge, I don't see them as being much different than "Enterprises". For basic connectivity (not all of the experimental stuff), if the R&E guys can do it, so can the "enterprises".

Every large Financial corp I have ever dealt with (all of them) already own their own *Metro* DWDM networks in various locations. The same can be said of many Hostpitals, Auto Manufacturers, Google, Yahoo, and Microsoft to name a few. Nortel, Adva, and Cisco sell lots of Metro DWDM in this space (probably more than to "carriers"). Arrangements with fiber providers vary everywhere from the Enterprise doing "everything themselves" (except fiber maint) to complete outsourcing to a service provider. It depends on what the company is comfortable with.

The Federal Government has already done this (ala GigBE) with Level 3 fiber in the long haul.

Wasn't there a ULH company specifically targeting enterprises with this idea a few years back?
Pete Baldwin 12/5/2012 | 2:59:04 AM
re: Enterprises: More Fiber in the Diet? We'd love to get more in-depth with one of these enterprise network projects. Anyone got suggestions for where to start? (Any enterprises want to volunteer?)
Pete Baldwin 12/5/2012 | 2:59:03 AM
re: Enterprises: More Fiber in the Diet? Thanks fiber_r_us.

Wasn't there a ULH company specifically targeting enterprises with this idea a few years back?

You mean Celion?

They didn't make it, though:

It would be interesting to hear their perspective on the market. Pawan Jaggi, if you're out there, give me a call: (415) 321-3750 x.28.
fiber_r_us 12/5/2012 | 2:59:01 AM
re: Enterprises: More Fiber in the Diet? Since practically everyone with a network of any size already does this with metro DWDM stuff (up to ~200km), is your question whether the idea is interesting over much longer (i.e. 1000s of km) networks like the R&E's NLR or the Fed's GiGBE?
OldPOTS 12/5/2012 | 2:59:01 AM
re: Enterprises: More Fiber in the Diet? I agree with fiber_r_us. Especially the market size comment (most Fortune 500 companys). That is how CSCO has grown.

I was recruited by a carrier vendor
to help carriers better understand their customer's needs, because we enterprise users (Fortune 10-100) better understood the business case and practical usage of these packet (ATM/FR/IP/Ethernet) networks. Think VBN, with a 'B' for Broadband.

Now when we enterprise networkers could, we had our own fiber installed when we could get right-of-way, or leased spare dark fiber from carriers or third party suppliers (like WilTel). Or else, if all else failed, we acquired (service &) capacity from carriers. Only a few carriers understood that not all of the network could be available to them and saw large $$ signs as they added up all our networks circuits. The original VPNs.

The leased dark and some lighted fiber usually worked for the core of the network. But remote sites, like Muleshoe TX or SE tip of LA, got carrier leased fiber and/or services - let the carrier service trucks roll. Carriers/third party suppliers had about half our $$ business. Mostly third party leases.

We only used bleading edge technology if it REALLY SAVED costs. We did converged packet networks back in 1987 - T1 & FastPacket. We used it a lot, then, but only when it really made sense from a reliability, flexibility, service and/or a real substantial ROI savings. We had no problem with removing some technolgies that failed to deliver (X.25). Or using TDM leased lines when packet equipment/line costs were significantly more expensive, even when allowing for growth.

We had our own intelligent multiplexers and fiber SONET/DWDM muxes for reliability control. That is why I think there is still a market for just providing fiber connectivity. While there is volume $$ in access, there is still equal $$ money in long haul interconnect.

The reason our approach was to own our own lines and equipment as much as possible, was for more reliable services (mesh fast recovery networks) to keep the business functioning and to save bottom line costs. The Fortune 20 CEO knew all our names and phone numbers as well as did the internal customers. We did a lot of spread sheets to determine the best business case for each business site, core network costs and requirement for convergance (mux/packet equipment).

New MBAs did comparative studies and customers graded us annually. Salary and bonuses were based on internal customer satisfaction and their reduced cost. And I'm now retired at age 60 and enjoying my sail boat.


PS - Scott, I would find a good enterprise user group (or periodical) for more market info.
Sign In