High-speed slowdown makes sense

Sept. 1, 2007

by Stephen M. Hardy

There will be plenty to talk about when the international optical communications community convenes in Berlin for the European Conference on Optical Communications. (Kurt Ruderman and I hope to see you there.) Certainly one topic of conversation should be the gradual warming of the European market. Signs of thaw can be found in a variety of areas. For example, the undersea market has picked up, as evidenced by Hibernia Atlantic’s plans to deploy a branching unit off its northern cable to provide connectivity to Iceland, as well as Tyco Telecommunications’ receipt of a contract from SEACOM for a new submarine network that will connect Europe with the Middle East, Africa, and India. The fibre-to-the-home market has begun to percolate as well, with the situation in France continuing to draw attention.

However, when talk turns purely to technology you can bet that high-speed networking, particularly the road to 100 Gbit/s transmission, will provide plenty to discuss. The IEEE High Speed Study Group recently agreed to add a complementary 40 Gbit/s amendment to its 100 Gbit/s Ethernet work. While the 40 Gbit/s effort is intended to focus on the needs of server-to-switch applications in the data centre-which means they’ll shoot for at least 100 m on OM3 multimode fibre, at least 10 m over a copper cable assembly, and at least 1 m over a backplane-the study group also agreed to ensure that whatever it comes up with is compatible with the ITU’s OTN/OTU3 rate.

The IEEE’s decision should be a boon for the makers of 40 Gbit/s technology. But the main reason the decision is a good one is because it recognizes service provider reality and maintains détente between the datacom and telecom worlds. The idea in some quarters that the growing ubiquity of Ethernet and packet traffic means that service providers should have to get on the 100 Gbit/s bandwagon or be left behind never made sense for several reasons.

Let’s start with current requirements. It’s true that increases in packet-based traffic drive higher network speeds; current vendors of 40 Gbit/s products concede that the demand for their technology currently rests on the 40 Gbit/s interfaces companies such as Cisco have installed on their large packet routers. You can certainly foresee that Cisco and other router vendors will add 100 Gbit/s interfaces to their boxes when they become available, thus sparking a demand for 100-Gigabit Ethernet. But we aren’t going to see 100 Gbit/s standards anytime soon-and even when we do, there’s no guarantee that the first generation of technology based on those standards will approach the price points carriers will demand.

Meanwhile, 40 Gbit/s interface technology is available today-and carriers have begun to use it. Service providers don’t increase transmission rates because they like installing cool new technology; the cool new technology always costs more than what they’re currently using. Carriers have moved and will continue to move to 40 Gbit/s transmission for the same reason they upgraded from 2.5 to 10 Gbit/s: Bandwidth demands won’t give them any choice, including the choice of waiting around at 10 Gbit/s for economical 100 Gbit/s transponders to appear.

And even when such transponders reach the field, 40 Gbit/s technology will still have a niche. For each route paved with the latest generation of dispersion-tolerant fibre there are several that currently struggle to support multiple 10 Gbit/s wavelengths. Many of these links will reach capacity exhaust just as surely as will the gleaming new 100-Gbit-capable backbones that are currently supporting 40 Gbit/s, and the next generation of 40 Gbit/s technology (provided the price points have come down sufficiently, including the necessary dispersion compensation) will provide the most economical and perhaps the only technologically feasible alternative to laying new fibre.

Finally, there’s a tendency in some quarters to assume that once a high-speed technology rolls out, its adoption is inevitable. Sometimes, that inevitability appears closer on the horizon than it really is. While 10 Gbit/s technology has been deployed for quite some time, plenty of 2.5 Gbit/s links remain. Similarly, while 100 Gbit/s speeds will someday be considered commonplace, carriers will still have 40 Gbit/s routes (not to mention 10 and 2.5 Gbit/s ones) that will enjoy long lifespans.

So there are plenty of reasons why 40-Gigabit Ethernet makes sense. The creation of such a standard continues the harmony between Ethernet technology and carrier requirements enjoyed at 10 Gbit/s. Given the stated targets for the main application, one can see the 40 Gbit/s Ethernet standards developers looking more closely at the 10GBase-LX4 and -LRM work than what has already been accomplished in the telco arena. Still, the opportunity exists for the High Speed Study Group to leverage at least some of the carriers’ experience.

Now if we can just figure out how to make it work.