Carrier-Grade Ethernet Faces Obstacles

Noting that almost all carrier services are becoming Internet Protocol-based, Hans-Ulrich Schoen, a Siemens AG vice president responsible for carrier products, said in his keynote address, "Ethernet is the natural language of IP." But first, he said, "Ethernet must become carrier-grade."

Schoen spoke at the All Star Network Access Workshop, sponsored by the ITU-T, the standardization arm of the International Telecommunication Union.

Many want to see a LAN connectivity technology progress to a carrier-class-service delivery technology, saying it will enable carriers to provide better flexibility through a far simpler, lower-cost interface. It will let users specify, for example, exactly how much bandwidth they want between the 10-Mbit/second and 1-Gbit/s range currently offered, proponents of carrier-grade Ethernet maintain.

Today's Ethernet lacks such carrier-grade features as quality-of-service, provisioning, fault-tolerant and self-healing specs, and what the telecom community calls "operations, administration and maintenance" features. "We know Ethernet is reliable within the LAN," Schoen said. "But we need to make it also reliable throughout the network." Moreover, he said, "when there is a problem of getting an access to an application in the network, we need to be able to pinpoint where the problem lies. Is the problem in the access, is it in the core or is it in the content platform?"

The ITU-T, IEEE 802.3, the Metropolitan Ethernet Forum and the Alliance For Telecommunications Industry Solutions are all jockeying to define different parts of the carrier-class Ethernet technology.

The ITU-T is jumping into the fray much more quickly than it has in the past. In the mid-'90s when ITU was far slower in standardization, the ATM Forum took over many service standardization issues.

The ITU said last week that it has recently approved new recommendations that describe Ethernet in a way that lets it be deployed in a carrier network over such technologies as asynchronous transfer mode (ATM), multiprotocol label switching (MPLS) and synchronous digital hierarchy (SDH).

Under the recommendations, enterprises can exist on one Ethernet-based LAN across a number of locations without having to interface with different technologies frame relay or ATM, for example between sites, the ITU said. That extends the flexibility of Ethernet over the WAN without having to add customer premises equipment or any other equipment at the carrier end.

But the ITU does not specify Ethernet transport network equipment as such. Nor do the latest ITU-T recommendations address the specific impact on the transport functions of connection monitoring in a connectionless layer network.

Bob Grow, chairman of the IEEE 802.3 Working Group, said that 10-Gbit Ethernet included carrier-specific functions approved in 2002 and that work has been ongoing to specify carrier-grade capabilities for deployment of Ethernet into the access market as part of P802.3ah, an amendment to IEEE Standard 802.3-2002.

Tasked to handle transport issues related to access stretching from the public network to the end user, the 802.3ah Ethernet in the First Mile group has just finished its draft for describing public Ethernet over twisted-pair copper.

"There will be always room for invention as Ethernet expands its scope and adds new capabilities," Grow said, stressing that IEEE 802.3 has always limited its work to media-access control and physical-layer functionality. "Therefore," he said, "higher-layer functionality work is done in other groups in consultation and cooperation with IEEE 802.3."

Indeed, industry forums have rushed to fill the service profile gaps in public-network Ethernet. For example, the Metro Ethernet Forum, a nonprofit organization to promote optical Ethernet in metro networks worldwide, has ratified the industry's first Ethernet service standard. "As Ethernet port costs are plummeting, it is a natural progress to put the same old Ethernet to new uses," said Metro founding chairman Ron Young. Metro has worked with groups like the MPLS/Frame Relay Alliance to describe how technologies like virtual private LAN services and transparent LAN services are handled over an Ethernet-IP-MPLS infrastructure.

Metro's focus, however, is squarely on Ethernet services, not on transport, said Richard Brand, director of network architecture at Nortel Networks. The industry still must look elsewhere for help in developing transport for carrier-grade Ethernet. A uniform management system is also needed for carrier Ethernet, he said.

Siemens has taken its case for carrier Ethernet to the Alliance For Telecommunications Industry Solutions, which adopts a business perspective in discussing what's needed to make Ethernet carrier grade.

Siemens promotes the idea that Ethernet, the dominant protocol in private networks, runs throughout the network, from the exchange's multiplexer to the end user, without the customary conversion to ATM. The goal is a "homogeneous, easy-to-manage network infrastructure, where costs can be cut because expensive network interfaces are eliminated," said Schoen.

That may be a tough sell in North America, where ATM dominates. But Schoen predicted that next-generation broadband networks will be based on Ethernet/VLAN and MPLS. "A combination of VLAN access and MPLS-based IP core is scalable and efficient," he said. "It does not require MPLS-aware nodes near to the customer location."

Additional reporting by Loring Wirbel.

This story courtesy of TechWeb.