by Meghan Fuller Hanna
By all accounts, reconfigurable optical add/drop multiplexer (ROADM) technology has become mainstream. First adopted by the cable multiple-system operators (MSOs), ROADMs now are finding applications in ILEC, CLEC, IOC, mobile operator, utility company, and research networks around the world. And these new applications are driving new features, like colourless operation, directional switching, higher levels of integration, and smaller, lower-cost options for the edge network. Says one industry insider, �We�re on the edge of a real application boom, and that will, of course, drive further innovation.�
ROADM capability is now considered a baseline feature for metro WDM deployments in many operators� eyes in part because the technology has matured, but also because prices have decreased to the point where the technology is really starting to prove in. �If you look at where a 10G transponder for a ROADM system used to be four years ago and where it is today, you�re probably talking about a 30% to 50% reduction,� reports Randy Eisenach, market development director at Fujitsu Network Communications (www.us.fujitsu.com). �We use that kind of as a benchmark for the overall level of reduction you can see on the chassis as a whole.�
But more than that, carriers are also beginning to appreciate the fact that ROADM technology, in general, is not as expensive as they initially thought it was, notes Ram Orenstein, associate vice president of product line management for ECI Telecom�s (www.ecitele.com) optics division. True, the cost of the ROADM component or line card remains higher than the fixed OADM, he says. But within the context of the overall network�including the chassis, the amplifiers, and the dispersion compensation�the ROADM accounts for just 5% to 10% of the total cost.
�Moreover, when you add ROADMs, in many cases, there�s an adverse effect on cost since they eliminate regenerators, which are costly,� Orenstein explains. �So actually, the more wavelengths you add to a typical ROADM network, the lower the total cost becomes compared to a traditional fixed WDM network. When we show this math to carriers, there is no question anymore that they should go for ROADM.�
Bill Kautz, manager of Tellabs� (www.tellabs.com) Global Portfolio Marketing Group, says he�s seen cases in which carriers lowered their capital costs by more than 50% with ROADM deployments. �That�s because you�re building a network in a fundamentally new way by only putting the electronics or the OEO electronics at the endpoints of the network, using the ROADM optical switching and optical pass-through at the intermediate nodes,� he explains. �That�s really starting to prove in.�
On the operational side, operators are also saving money, as they no longer need to send technicians into the field to provision a new service at these intermediate nodes. And they can carry fewer spare parts because a single ROADM covers the entire C-band versus a fixed OADM, which requires multiple spare parts.
For his part, Orenstein believes the ROADM market �is booming to such an extent, it�s becoming...I wouldn�t say a commodity, but I would say every other optics RFP we answer is with a ROADM.�
To become a commodity, ROADM technology would have to migrate to the area of the network that supports the highest volumes: the edge. And all sources interviewed for this story say that�s exactly where the ROADM is headed next.
Today, ROADMs are primarily deployed in metro core and regional networks, but system vendors say their customers are now interested in deploying ROADMs both deeper into the core and, especially, out to the edge. Edge ROADMs are designed to give operators the key benefits of the core ROADM�including single-channel granularity and auto-power balancing�in a smaller, less expensive chassis.
However, to achieve the kind of price points required for volume deployment at the edge, system vendors will have to make some functional tradeoffs. For example, the typical edge ROADM will limit the number of wavelengths an operator can access. �When you�re talking about edge ROADMs,� says Paul Morkel, ADVA Optical Networking�s (www.advaoptical.com) senior director of business management, carrier WDM, �a 25% add/drop capability is more often than not an acceptable configuration, meaning you can drop 25% of the total number of wavelengths coming through the site.�
In a 40-channel system, an operator would be allowed to drop 10 wavelengths. But herein lay the challenge for vendors developing edge ROADMs: Network operators do not want to be restricted to pre-assigned wavelengths. They want to be able to drop any 10 wavelengths on that 40-channel system. �That�s been part of the challenge,� Morkel admits, �being able to provide that level of functionality at the right price points for the edge.�
System vendors also see the need for higher-degree ROADM components and subsystems. �If eight- or ten-degree ROADMs are popular today in the core, we expect to see 20-degree ROADMs going forward,� says Orenstein. �Not so much because we expect to see nodes with 20 degrees of fibres, but because bigger carriers value the combination of higher nodal degree with a high number of colourless ports for local add/drop. To do that, you need a WSS with a higher port count than the currently available device,� he says.
Most ROADMs today are coloured, meaning particular ports are still associated with particular wavelengths. Operators must manually connect the fibre at the endpoints for service turn-up. As such, a number of operational benefits could be attained from removing this restriction. In a colourless configuration, any wavelength could be connected into any client port on the ROADM, enabling operators to prefibre transponders.
�If you�re trying to do fully software-provisioned services whereby you don�t even have to touch the endpoints when you turn up the service, you could have it all prefibred,� explains ADVA�s Morkel. �Then you set the wavelength at the time you need the service, and you decide which wavelength you want to use at that time.� In that case, he says, you really need to have colourless capabilities.
�And that�s where a 20-degree wavelength selective switch is very interesting,� he continues. �Because the wavelength-selective switch ports themselves are colourless, you can use those not only for network connections but also for the client connections.�
The folks at Cisco Systems (www.cisco.com) also argue strongly for the need to increase colourless port density. �Today, you can really only buy it in eight-port chunks,� laments Russ Esmacher, manager of technical marketing for Cisco�s Optical Transport Business Unit. �If I want to do a full, 40-channel colourless add/drop, I need to have five of these things put together, and that�s big and expensive. We need to increase colourless port density.�
Hand in hand with colourless capability is the need for directional or directionless switching, which Cisco claims to have been the first to develop in both hardware and software. Directionless switching is the ability to connect any client port or interface to any network port in a two-degree or multidegree ROADM. For his part, Esmacher believes that a colourless configuration without directionless switching is �completely useless. Great, I have the ability to have colourless, agnostic ports. Too bad I�m assigned to direction north. I need that to go out to the west, and I need to change that without moving any cables,� he says.
System vendors like Cisco are now relying on the component vendors to deliver the requisite port density to enable these new features. �We�re waiting for it,� reports Esmacher. �We have the architecture built and the software going for it. We just need better density. When we sit down and talk to our [component] vendors, these are the things we ask them about, very strongly.�
There is good news, says ADVA�s Morkel: �Wavelength-selective switching is a platform that is capable of doing more than has been implemented in the first generation of ROADM.�
On the client side, network operators are also beginning to push for increased integration of what were, traditionally, separate network elements, and this has given birth to a new class of equipment known as the packet optical network platform (PONP) or, as popularised by Verizon in its recent RFP, the packet optical transport platform (P-OTP).
PONPs are architected to support SONET/SDH, Ethernet, and WDM�including ROADM and optical switching�with equal deftness in the same platform. While the system vendors themselves are in various stages of PONP development (or, in the case of Cisco, eschewing it altogether), most agree that their customers are, at the very least, intrigued by this level of integration.
Eisenach says Verizon, which obviously has a lot of leverage and purchasing power in North America, is �wholly embracing the concept of TDM aggregation, packet aggregation, and ROADM integrated in a single chassis.� Others are still trying to determine how they will internally manage and administer the convergence of two capabilities�TDM and packet�that were historically supported on two separate platforms and networks.
�I think they are just working through those issues,� Eisenach says, �looking at how they make that happen, how they implement it. But I think they understand it�s the right thing to do because of the cost savings.�
Tellabs� Kautz agrees. �Everyone has looked at ROADM as an operational expense savings, and it is. But it actually turns out to be the integration of ROADM and Ethernet and SONET/SDH that is really a capex savings,� he maintains.
Meghan Fuller Hanna is senior at Lightwave.