40G a big bust? What does the future hold for ultrahigh-capacity transport?

Telecommunications Americas, August, 2003 by Sam Masud

Were optical equipment vendors trying to sell carriers a white elephant with all the talk about 40 Gbps? In 1999, at both the SUPERCOMM show and the ITU's Telecom World event, 40G was the hot topic. Hanan Anis, CTO of Ottawa, Canada-based start-up Ceyba but who previously was part of Nortel's 40G effort, recalls that back in '97 there was even talk of 80- and 100-Gbps wavelengths. For its part, Nortel acknowledges being behind the public push for 40G and concedes it miscalculated market demand.

Rob Keates, director of optical networks marketing for Nortel, says the company still has the capability to bring 40G to the market but at this time sees no demand for this technology on either the transport or the switching equipment. "The drive to 40G was predicated to a degree upon growth that would occur at the wholesale level, but because the number of carriers has been reduced dramatically there's a lot less wholesale business out there," says Keates.

Infrastructure capex and opex are the twin devils for carriers and vendors like Nortel and Ceyba believe that the demand for 40G will be there when it can meet both those challenges. The jump from 2.5- to 10-Gbps channels was driven by price because carriers could get four times as much capacity at two-and-a-half times the cost of 2.5-Gbps channels. The same math will also have to apply for carriers to make the transition from 10G to 40G. On the other hand, it might be easier for carriers to manage a few 40G channels than four times as many 10G channels.

Some Possibilities

There are two areas where Nortel continues to see a potential for 40G:

* specific high-traffic routes where, if the cost proves in, it might make sense to move up to 40G channels; and

* in so-called express routes where a carrier would want to do an optical by-pass.

Heinz Zallman, director of portfolio management and strategies for Lucent's optical networking group, agrees with Nortel and notes that the Northeastern corridor as well as the West Coast could become the initial candidates for implementing 40G networks. For, according to Zallman, while a carrier might--on average--have overcapacity in the network, there are routes where congestion could become a real concern as installed DWDM systems reach capacity. Thus building 40G networks along the two coasts might also entail deploying express routes to connect those networks.

Having an express route means no expensive OEO (optical-electrical-optical) conversions, and even where such conversions might be necessary, 40G proponents note that it would take four times fewer transponders to accomplish these conversions than would be needed for 10G wavelengths. The advantage of fewer transponders, however, would be offset by the fact that the reach of the 40G optical signal would be impacted by issues such as PMD (polarization mode dispersion) and PDL (polarization-dependent loss), both of which degrade the signal and, unless addressed, require frequent opto-electronic regeneration of the signal. Thus carriers might be unlikely to adopt 40G unless 40G networks have the same distance reach as 10G networks and/or show some real cost benefits over 10G.

Ceyba hopes to establish itself in the transparent optical long-haul network with a platform that supports both 10G and 40G channels in the same band, with its 10G technology having' a reach of up to 4000 km and 40G wavelengths a 1500-km reach. "We're acutely aware that nobody's deployed anything (i.e., optical gear) of significance in the past year," says Anis. "But every customer we've talked to wants to see 40G demonstrated along with 10G on our platform." Anis says 40G has been a victim of a double-whammy: "People have a lot of 10G in inventory and so they've dropped the price [of 10G] below cost. Secondly, there's no economy of scale to bring down prices for 40G."

Although Ceyba has declined to identify any customers, a demo or trial of 40G is apparently as far as carriers are prepared to go right now, at least in North America. For while Lucent supports 40G on its LambdaExtreme DWDM system and has had trials of the system, at least at this time it claims to have only one trial underway, and that too is overseas.

Who's First With 40G?

If the outlook for 400 is lukewarm at best, chip vendors, such as start-ups Ample Communications and Teradiant Networks, are bullish about 40G. The latter makes a 40G packet engine that has the flexibility of network processors and performance of ASICs and also a traffic manager chip--devices that reside between the switch fabric and the frame processor. Ample, on the other hand, makes a 40G frame processor that sits between the optical module and the network processor. The two vendors are very much in the early stage of 40G, with Teradiant just starting to send out sample quantities and Ample noting that thus far only test system vendors have received its product.

However, both vendors predict that the router equipment companies and manufacturers of multi-service switches will likely be the early, adopters of 40G. "We think 40G will show up first on routers to interconnect them within the CO or the POP, because a big problem is all of the cables and fibers going between these boxes," says Marek Tlalka, Ample's vice president of marketing. "The reason why 40G didn't show this year at all is because you couldn't buy an optical module. All of the components that go into the optical module you have to build yourself, which is very difficult. Only a very few [companies] like Nortel, which has years of experience, can do it."