I have to admit to raising an eyebrow when I first heard about the new Carrier Ethernet initiative launched by the MEF. (MEF, of course, stands for Metro Ethernet Forum-but now that the organization has decreed that Ethernet isn’t just for the metro anymore as far as carriers are concerned, it has a problem with its name. The current solution appears to be to just use the acronym.) As MEF president Nan Chen and Ethernet pioneer Bob Metcalfe spun their vision of Ethernet as the ubiquitous and primary protocol over which all services were going to run in all parts of a carrier’s network, eventually flowing directly over glass, my skepticism sprung from the feeling of déjà vu their tale engendered. Because of course we’ve heard all that before, haven’t we?
Just a few years ago, Internet protocol evangelists wandered out of the desert with the same message. As Internet traffic came to dominate carrier networks, the inefficient way in which SONET/SDH handled packet-based streams would become manifest and abhorrent in the sight of carriers, which would then convert to the IP religion. The advent of MPLS enabled IP to support resiliency and reliability comparable to SONET/SDH; network planners would look upon the IP-over-glass network and it would be good.
If you replace “IP” in that last paragraph with “Ethernet” and “MPLS” with a few new IEEE and MEF initiatives, you’ll pretty much have the story I heard from Nan and Bob. (We’ll examine the case for Carrier Ethernet-with significantly less flippancy-in our Case by Case column next month.) Of course, if you go back a few years before the advent of the Internet, some ATM acolytes also proclaimed SONET/SDH as corrupt and their protocol as the path to data-friendly network enlightenment.
In other words, we’ve had several replacements for SONET/SDH offered to carriers, and none of them has caught on, at least not in the sense of eclipsing the current optical-network foundation. That thought leads me to wonder why the latest pretenders to the SONET/SDH throne should be any more successful than IP and ATM. Yes, the technical aspects sound plausible-but so did those of IP. True, some carriers have begun to implement Ethernet beyond the metro. But some carriers did the same with IP, even in the long-haul. The best known of these, Enron, is no longer with us.
Now it should be understood that while SONET/SDH won’t depart from carrier networks anytime soon, its days do appear to be numbered. My wise-guy attitude aside, the criticisms I described have merit. While the protocol can handle packet traffic, it does so inefficiently. As such traffic comes to dominate carrier networks, particularly as voice moves to IP, it makes sense that carrier networks would evolve toward optimized packet handling, particularly if SONET/SDH-like reliability could be assured.
So I believe some protocol will replace SONET/SDH (pretty far) down the road. But will that protocol be Ethernet? I recall participating in a panel at a trade show in Paris some years ago; the topic was something along the lines of “the future of optical networks.” The industry was at the height of its IP hysteria, and a vendor spokesman made the case for an IP world. I’m not sure why-probably just to create controversy, certainly not because I’m any kind of visionary-I suggested that Ethernet might challenge IP in the future. The spokesman pooh-poohed the notion, since IP could now offer carrier class operation and Ethernet could not.
I didn’t have the chance to point out that IP couldn’t support carrier class operation, either, before MPLS, and there was nothing to prevent the Ethernet community from developing its own version-which they have now done. And with that achievement reached, the Ethernet proponents see no reason why their proposals shouldn’t carry the day, just as their IP brethren believed before them.
But it’s that “always another option down the road” experience of mine that leads me to hesitate before jumping on the Ethernet bandwagon. Just as the IP folks never saw Ethernet coming (and most carriers didn’t see IP coming), it’s sometimes tough to see beyond the glow of your protocol religion into the future.
I’ll give the Ethernet proponents this: There doesn’t appear to be an existing potential challenger that is likely to get a carrier class upgrade in the near future. Frame Relay isn’t going to cut it; Fibre Channel is too much of a niche protocol.
So maybe Ethernet is the long-awaited packet-network messiah. But I’m going to remain agnostic until the specter of false prophets recedes.
Stephen M. Hardy
Editorial Director & Associate Publisher