As reconfigurable optical add/drop multiplexers, or ROADMs, evolve, each new generation of the technology has proved more prevalent and important to carriers, becoming a must for any metro optical equipment vendor portfolio.

Just a few years ago, they were designed for long-haul core networks. Then they were applied to metro networks. And then they became more powerful, with the help of wavelength-selective switching (WSS), which allowed the remote switching of wavelengths among multiple rings (Telephony, April 5, 2005). Vendors are still following that trend; Tropic Networks (whose ROADMs are sold by Alcatel) and Ciena, for example, both announced WSS gear last month.

So what's the next step in this technology's evolution? One optical equipment vendor, Lambda Optical Systems, might argue that the next step is the one it recently took: ROADMs based on advanced telecommunications computing architecture (ATCA), a modular approach that lets vendors create gear from interchangeable subsystems. System vendors seem to have mixed views of the notion so far — Ciena said such an approach might make it harder for vendors to differentiate. Meriton Networks said it's watching the space closely — but carriers seem unconcerned with it.

“I have not seen [ATCA] as a requirement from the bigger carriers in their [requests for proposals],” said Ron Kline, research director for research analysis firm Ovum-RHK. “I'm not sure the big vendors are focused there at all.”

The historic path of network technologies originally introduced in the core is to slide downstream to metro, then edge, then access networks as its price declines and as network capacity grows. ROADMs already have slid from core to metro networks. Could edge networks be the next stop?

At last month's Globalcomm 2006 trade show, a new start-up, Nistica, emerged from stealth mode pledging to be the vanguard of the edge ROADM movement. The company offers three different sizes of edge ROADM subsystems that, between them, can add or drop between one and 16 wavelengths. But Nistica's founders expect their most popular product to be the mid-sized box, which offers the “just right” porridge of four to eight 10 Gb/s wavelengths. About 10 system vendors already are evaluating Nistica's gear, which will be generally available in the fourth quarter and could be introduced to carriers next spring, the company said.

Carriers will need this gear, according to Nistica, largely because of the flood of bandwidth in access networks resulting from telco video services and the pressure it puts on edge gear.

“There's this huge gap between the [passive optical network] area and the metro core,” said Ashish Vengsarkar, Nistica's CEO. “Bandwidth requirements are forcing [carriers] to add wavelengths. Once they go past 10 Gb/s, instead of pulling new fiber, they want edge boxes that help deliver wavelengths.”

ROADMs are nothing new to Vengsarkar and his team. He, along with Nistica's chief technology officer and its chief financial officer, helped found Photuris, a start-up vendor that claimed an early Verizon metro ROADM deployment (for a private customer network, a university) before going out of business in 2004 when the company, running out of cash, failed to win a competitive SBC Communications contract. Having raised more than $105 million in venture capital, Photuris sold out to system vendor Mahi Networks for $1.8 million. (And Meriton Networks acquired Mahi last year.)

But the notion of an edge ROADM might be naturally strange to some because the benefits of ROADMs — the remote, automated reconfiguration of wavelengths from one path to another — may seem inapplicable to edge networks for at least a few reasons. For one thing, in edge networks, traffic moves much more predictably than in a metro ring or mesh, usually from access networks to aggregation points, often in a hub-and-spoke pattern. In particular, that might preclude a need for the multi-degree wavelength switching offered by WSS, but it also calls into question the need for automated reconfigurability at the edge.

“Reconfigurability is not as critical at the edge,” said Glenn Thurston, vice president of marketing for BTI Photonics. “At the edge, it's about adding wavelengths. Once a wavelength is up, it tends to stay up for three years. Today, fixed [WDM] is still the way to go.”

ROADMs still have a place at the edge, Thurston said, but their value proposition is different there than it is elsewhere in the network. For example, an often-repeated selling point of core and metro ROADMs is the ability to avoid a “truck roll” (or onsite visit) every time a wavelength needs to be reconfigured. But at the edge, you can't avoid a truck roll, Thurston said; new links will need to be physically connected no matter what. It's the additional engineering required by those changes — the changes in power supply, etc., that add time to that truck roll — which edge ROADMs can help alleviate.

“The real value is, if you do move traffic around, it's a pain,” Thurston said. “A ROADM simplifies that complexity.”

In May, BTI introduced a multiplexer designed for the single-digit channel counts of edge networks that remotely monitors and manages power levels. It started shipping in June.

“We've got [CLECs] and large companies ordering that today,” Thurston said. “It's saving them a lot of time in terms of sending a guy out to manually adjust the power as you're adding wavelengths.”

More ease and automation in the process — no matter what kind of product delivers it — will allow carriers to shorten wavelength provisioning times, a difference they will use to compete against one another. “The guy who can deliver a service in a day can beat the guy who takes a week,” said Bill Gartner, Meriton Networks' chief operating officer.

Though Nistica points toward residential video service as a driver for edge ROADMs, B&C Consulting President Clif Holliday doesn't see a need for edge ROADMs in triple-play networks. Consumers might make use of 50 Mb/s or even 100 Mb/s to each home, “but it's nowhere near a wavelength,” he said. “I don't see it on the residential side for years.”

Holliday does see a need for edge ROADMs in enterprise applications as the market for wavelength services heats up, driven by corporate concerns over disaster protection and business continuity. “When you get into that kind of market, a ROADM is a natural for that last main network node before you hit the customer,” he said. “It would make it so easy to drop a wavelength.”

However, regardless of the market or application, the age of the widely deployed edge ROADM is not here yet, Ovum-RHK's Kline said. “I haven't heard of any service providers knocking down the door to get mini-ROADMs. Although I can imagine that, once telco and IP video take off in a big way, the application will be there.”

To be deployed in edge networks, the price of ROADMs will have to come down, and they are. Today's ROADMs are roughly 80% what they cost a year ago, Holliday said. But they're still not cheap. A 32- or 40-channel ROADM with 10 Gb/s interfaces might set you back close to $1 million today.

On a component level, vendors might need to “knock a zero off the price” to make edge deployment feasible, BTI's Thurston said.

But as big carriers — and cable operators alike — are forced to deploy thousands of ROADMs nationwide to support big-bandwidth video services, the price should drop sharply, particularly among the high-end WSS ROADMs that are likely to be deployed in the most volume. (AT&T, for example, has vowed to deploy 400 ROADMs nationwide this year to accommodate video service.) Nistica claims its edge ROADMs, which are WSS-based, will be 80% cheaper than the modern metro kind.

At the edge, a product's upfront cost is a much more important consideration than it is in metro and core networks. In fact, it may dictate solely where ROADMs go next. No matter what the other functional benefits of the technology are, the timing and even the feasibility of their arrival at the edge, Meriton's Gartner said, “will be driven entirely by the cost function.”