Less is more for next-gen ROADMs
By Stephen Hardy
Colorless, directionless, contentionless, and gridless features are on the horizon. Yet not every carrier is clamoring for them.
First-generation reconfigurable optical add/drop multiplexers (ROADMs) are not always as reconfigurable as some service providers would want. The ideal platform would enable a carrier to add or drop (or express) any wavelength at any frequency in any direction, easily and without a truck roll.
The attributes that would contribute toward such nirvana are commonly described as:
- colorless: the ability to accommodate any wavelength
- directionless: the ability to enter and exit the node from any direction
- contentionless: in which the switch architecture prevents wavelengths from bumping into each other during path reconfiguration
- gridless: the ability to change channel plans and wavelength spacings to accommodate guard bands or changes in modulation format, increase channel density, etc.
Systems suppliers appear universal in their desire to make available all of these features. Yet conversations with vendors and carriers reveal that the two parties aren't necessarily in harmony when it comes to how these attributes would be used, where they would be used, or whether they're even necessary.
Gridless technology is a more recent target, but systems houses and their subsystem suppliers have been working on colorless, directionless, and contentionless (CDC) capabilities for several years. Yet despite the claims of some vendors, carrier sources such as Glenn Wellbrock, director optical transport network–architecture and design at Verizon, say that ROADMs with all three features (or at least those designed for the long-haul, packet-optical, transport-friendly applications where Wellbrock would like to deploy them) are not yet available.
"The end points are still very static," Wellbrock reports. "Per-port basis is still colored. If you want to go from Channel 1 to Channel 2, you have to move it–or if you want to go from direction one to direction two, you have to physically move to another WSS [wavelength selective switch] at the access/egress points. Now at intermediate locations, yes, you can come in and go out in different directions. It has to remain the same wavelength; you can't change wavelengths."
|Carriers have deployed ROADMs, but they're not as elegant as many service providers would like. (Fujitsu photo)|
That said, Wellbrock thinks systems developers are getting close to enabling CDC capabilities. "I believe we'll see early prototypes this year. I believe we'll probably see products next year," he offers.
Mark Feuer and Sheri Woodward, principal members, technical staff at AT&T Labs, agree that progress is being made, at least in colorless and directionless capabilities. "The key is getting the details right and getting the cost points right," Feuer says. "And speaking just as a researcher, I think that we can expect to see major carriers deploying colorless and non-directional pretty soon."
System developers agree that the contentionless aspect of CDC is the most difficult to perfect. "I'm not convinced that switching matrices of the sorts of sizes you'd need are easily available," says Robert Keys, chief architect, optical, at BTI Systems, when discussing the MEMS-based switching subsystems that might be used to create a contentionless capability.
The other path to a contentionless architecture would use WSSs and arrayed waveguide gratings (AWGs). But hurdles still reside here as well, particularly when you take into account the large numbers of wavelengths major hubs would have to accommodate. "Just simple things like 96-channel AWGs–they don't really exist today," Keys explains. "Then you're doing 48-channel AWGs and interleavers. And suddenly the complexity of that node from an optical construction [standpoint] really starts to get up there quite significantly."
|WSSs–as illustrated by this photo from 2008, showing a 1x2 device from Xtellus (now part of Oclaro)–have been available for several years.|
Complicating the matter is a lack of consistency in terms of customer requirements. While Wellbrock appears interested in contentionless capabilities, Feuer and Woodward are much less enthusiastic. AT&T has studied various options and discovered that if you put a client-side switch between the ROADM transponders and the routers or whatever box is accepting and feeding wavelengths into the ROADM, you can solve virtually any contention problems within the ROADM itself.
"That switch does a lot more than merely help you manage contention within the ROADM. It provides the ability to do bridge and rolls if you want to, for example, groom your network traffic," Woodward adds. "It also provides the ability to have 1:N transponder protection."
Summing these factors, the benefits of having such a switch in place outweigh the advantages of adding contentionless capabilities to a ROADM, Feuer and Woodward believe. "If you have this switching on the client part of the node, the small bit of contention within the ROADM part might become insignificant," Woodward concludes. "And so the jury is still out on whether there will ever be a contention-free architecture that pays for itself."
The jury also appears to be out when it comes to gridless capabilities, at least in terms of the value of the feature and how it might be used.
Wellbrock sees the benefit of grid-less technology in accommodating the channel-width requirements of whatever exotic modulation formats support 400 Gbps, 1 Tbps, and future jumps in data rate. "What we've been asking for is 25-GHz granularity," he says. "The primary benefit that we're looking for is to go from 50 [GHz on the ITU grid] as we start to 125, 150, or 300 [GHz] based on the assumption that we're not going to have a lot of different bit rates."
However, gridless capabilities might create some problems as well, the AT&T researchers believe. "If you're not careful as traffic grows, you could end up stranding bandwidth," Woodward explains.
Yet, such concerns should not overshadow the benefits gridless, Keys and Jim Theodoras, director of technical marketing at ROADM vendor ADVA Optical Networking, assert.
"Having gridless lets you save guard bands, but it also lets you play with some of the edges of the bandwidth," Theodoras explains. "So as you go through multiple stages–and we have one network where there are 18 ROADM nodes to go through–the edges of a band is where your performance degradation occurs." Having grid flexibility would improve performance in such applications, Theodoras says.
"Where flexibility adds a lot of value is where you've got a lot of change in your networks," Keys adds. "The rate of change of services in that sort of [long-haul] network, the churn, is much slower than, say, in an edge metro network where there's a lot of change, a lot of churn, a lot of services being added."
The belief that gridless capabilities can help accommodate such churn will lead BTI Systems to have some form of the capability available in the very near future.
In fact, it appears that CDC and gridless ROADM capabilities are just around the corner. But the ubiquity with which carriers will deploy these features may require some reconfiguration, as well.