Is there somewhere where one may get a short course on the players here?
On Cambrian's website, click on "About WDM" for a list of articles. I'll post one that mentions most of the key vendors.
cambriansys.com
Ciena's troubles are a good reminder that sins of omission are as damning as commission.
Oh, what a tangled web we weave. . .
Later --
Pat
September 15, 1997
Transmission special feature
Toward an optical layer
Photonic switching will be key to developing an all-optical network
STEVEN TITCH
Not long ago, the only place to hear energetic discussions about photonic switching, dispersion infusion and rolloff was the Bell Labs employee cafeteria.
These days, however, all manner of terms optical are flowing eloquently from marketer's lips, like so many sales numbers from their Excel spreadsheets.
Some of the more dazzling products that have been displayed recently or that are slated for the upcoming National Fiber Optic Engineers Conference are at best prototypes. They indicate, however, that vendors are plotting out routes to the "optical layer," the next stage in large-scale networking beyond the Sonet hierarchies and wave division multiplexing (WDM) schemes now being offered.
For vendors such as Lucent Technologies, Northern Telecom and particularly Alcatel, it's a chance to remind the market of their long and successful histories as innovators in fiber optics. For others, including Ericsson, Ciena, Cambrian Systems and Hitachi Telecom, it is an opportunity to create or enhance their identities as fresher, newer voices behind transmission platforms that transcend traditional time division multiplexed (TDM) models.
Not only are companies such as these introducing or enhancing wave division multiplexers and terminals, but at least four have taken tentative steps toward photonic switching by highlighting rudimentary optical cross-connect switches.
Explosive growth
Although much attention has focused on the effect of the growth in Internet usage on local exchange carrier network capacity, interexchange carriers are under just as much pressure. In fact, the IXC networks are handling the immense amount
of nationwide--and global--Internet protocol traffic generated by Internet users.
Patrick Nettles, president and chief executive officer at Ciena, notes that the amount of data transversing the public switched network is growing 10 times per year. That, he says, translates to 2% a day. "That growth rate has huge implications for the network," he says. Meanwhile, he adds that in just a few years, Internet traffic will incorporate both data and voice.
"Demand for bandwidth exceeds Moore's Law," says Eric Sumner Jr., architect and chief technical officer at Lucent, referring to the often-quoted statistic that computer processing power doubles every 18 months to two years. "It's going to change the way we architect the network."
Those sizing up the situation are unanimous. "The Internet is putting demands on operators to provide bandwidth," says Jose Maria Daza, vice president of business development for Ericsson. Carriers, he notes, are beginning to move toward WDM architectures to optimize the use of their networks. "This is applicable for metro and long-distance networks," he says.
Much like a prism, WDM breaks down a single lightwave signal into several wavelengths, visible as colors. Current systems now can multiplex up to eight or 16 wavelengths, although several vendors have announced 32-wavelength multiplexers. WDM typically is used to get more capacity out of OC-48 (2.4 Gb/s) Sonet, the speed at which most IXCs currently operate their backbones. LECs still mostly use OC-12 (622 Mb/s), but because of traffic projections, they are under tremendous pressure to upgrade (Figure 1).
As an alternative to WDM, carriers also have the option of going one more step up the Sonet hierarchy to OC-192 (10 Gb/s). But OC-192 requires a better grade of fiber and, often, the infusion of additional dispersion--something that fiber manufacturers have been working to limit in each new generation of optical fiber. WDM also requires some additional dispersion because of spurious signals at different wavelengths.
Although talk about the possibility of OC-768 (40 Gb/s) continues, the general consensus is that TDM architectures such as Sonet are reaching their limit in terms of cost-efficiency and the next level of the network will involve greater use of WDM and optical switching (Figure 2).
Sonet multiplexers ultimately could be displaced by other terminal equipment that interfaces to OC-48, Ciena's Nettles postulates. Sonet, in addition to being a TDM hierarchy, provides a certain degree of management functionality, he notes. There is no reason why, in a gigabit world, that functionality couldn't be handled in the photonic layer. Sonet, above all, is an interface, not a multiplexing requirement, he says.
Some say the management functionality may be handled by an optical cross-connect switch. Alcatel, Lucent, Hitachi and NEC all have discussed some version of such a switch, although all readily admit that the models did not incorporate the features or robustness that would be expected from a commercial optical switch.
Still, they have made a start. Alcatel and Lucent talk about using photonics to build on infrastructure they already supply to customers. Hitachi and NEC, companies that have had trouble penetrating the U.S. public network market, hope to use photonics to build a presence. Hitachi's major victory thus far has been the sale of a point-to-point OC-192 system to MCI, which is using it to route traffic between Chicago and St. Louis.
IP-to-optics?
One of the more provocative notions has been that photonics could completely replace the Sonet TDM layer.
Phillip McCall, vice president of sales and marketing for transmission systems at Hitachi Telecom (USA), boasts that his company has found eight alternative network architectures to the current IP-to-asynchronous transfer mode, ATM-to-Sonet, Sonet-to-photonics approach. The most radical of these suggests going directly from IP on the desktop to an optical connection.
"The Sonet layer is somewhat inherently inefficient," he says, citing the standard use of one-to-one redundancy. While there is a certain level of comfort that comes with full backup, increasing demands for capacity now have carriers eyeing that spare capacity hungrily. More flexible redundancy in the optical layers, often cited as n-to-1, are becoming more desirable, McCall says.
McCall also points out the extra fiber quality and higher level of optoelectronics required by OC-192 and higher-speed Sonet, in comparison with the near commodity market that exists for OC-3, OC-12 and OC-48 today. It's also true that OC-192 has not been completely standardized.
"There's a need now for routing optical signals without having to come down to electronics," McCall says. Hitachi's optical cross-connect is the first step, he suggests. Intended primarily for restoration applications, Hitachi's device is now in a field trial with MCI in Dallas (Telephony, Aug. 25).
Some Sonet elements could run unprotected, says McCall, driving the ratio of live and backup circuits to 2-to-1. "We're being pushed very hard," he adds.
Hitachi's vision of IP-to-optics gained some support from others. John Ryan, principal with Ryan Hankin Kent, a consulting firm that Hitachi has retained, can't see why not. Running "naked IP" to the optical sublayer already means two boxes instead of four. Why "slice and dice" IP into 53-byte cells? he asks.
Gerry Butters, North American president at Lucent Technologies, thinks the idea is feasible, making the company one of the few large Sonet vendors that believes that some day the Sonet layer could be replaced. A carrier network goal should be to get from IP to light as quickly and cheaply as possible. This might mean evaluating product life cycles in the transmission infrastructure, he says.
Lucent recently exhibited a 16 x 16 optical cross-connect switch, which will have add/drop multiplexing (ADM) capability by year's end, Butters says.
Still, getting from the Sonet layer to optical layer is still costly, as Ciena's Nettles points out. OC-192 and OC-192 with WDM will become increasingly attractive.
There's no limit to TDM, notes Ericsson's Daza, who says that TDM and WDM will complement each other--although for the moment WDM is more cost-
effective.
A hybrid approach
Most believe the Sonet layer will be around for some time to come--at least 10 to 20 years.
"Everyone can see the vision of optical networks," says Tom McDermott, chief technology officer at Alcatel Network Systems. "Many are still struggling with it."
Alcatel considers its optical cross-connect switch lab grade now, he says, with optical ADM technology within reach in 12 to 18 months. Practical optical cross-connects are at least five years away, says Tim Krause, director-product marketing business development in Alcatel's optical networks unit.
WDM, however, will come of age in 1997, both McDermott and Krause say. "Equipment is coming up the curve," particularly in the areas of filters and precision layers, says Krause.
In addition to the capacity gains they can provide, another appealing aspect of Sonet and optical transmission is reliability.
As IXCs reach commodity pricing, competition is shifting to how well downtime can be avoided and, in the event of service disruption, how fast service can be restored. Everything is that much more acute when the driver is data traffic, especially commercial transactions over private networks.
Recovery demands have gone from days to hours. Now they must go from hours to minutes, says Nettles. "Density relies on reliability," he says. As the cost of bandwidth plummets, traffic will increase, even though the number of subscribers in a given area may remain constant.
Cambrian Systems, which is planning to introduce a 32-channel dense WDM system called OPTera at NFOEC, sees hybridization between electrical and optical cross-connect switches. Don Smith, president of Kanata, Ontario-based Cambrian, envisions a network architecture in which ATM, IP and other protocols directly interface with the optical layer along with Sonet and plesiochronous switches. A key part of that scheme is the migration of management and protection into wavelengths within the WDM system.
"All kinds of edge devices are delivering OC carriers. But Sonet does protection switching," says Smith. The goal of the Cambrian architecture is to bring that protection switching down to the OPTera, he says. "The issue is, how do you take the wavelengths and turn them into a manageable tool?"
Cambrian's push is the interoffice network in the local exchange. The OPTera concept is a 80 Gb/s WDM ring of up to 20 miles without amplification. It can accommodate up to eight remotes and one hub.
Looking to the future, most see more functionality moving into the photonic area. The need for an optical cross-connect is likely to emerge, but there are questions as to how fast vendors can get such switches to scale. There is also no getting away from the management aspects. Some say that handling information at higher optical levels will mean a layered structure--perhaps an in-band/out-of-band format.
But vendors admit that high-level photonic solutions could be some ways off. Right now, says Nettles, Ciena's focus is on real customer opportunities. "We don't want to talk about problems that aren't of a tangible nature," he says. "We look for customer needs. We start with a customer problem and look for a solution."
Contact Editorial Director Steven Titch.>>>> |
